A Triadic Intervention for Adolescent Sexual Health: A Randomized Clinical Trial

The current investigation is a formal efficacy trial of the Families Talking Together (FTT) triadic clinic intervention. We conducted a parallel RCT in which mother-adolescent dyads were randomly assigned to a control group that received no intervention, an active control group that received a nutrition-focused intervention, or an experimental group that received the FTT intervention. Data revealing preliminary efficacy and lessons learned from a small pilot study³³  informed the methodology for the current investigation.

The research took place in a community-based health clinic located in an urban context with elevated incidence of teenaged pregnancy, STIs, and HIV among adolescents.^34,35  The clinic is a federally qualified health center³⁶  that provides primary and preventive care to underserved individuals regardless of income or health insurance status.³⁷ 

Mother-adolescent dyads were recruited in a health care clinic during a non–acute care visit (eg, annual physical examination and routine follow-up). Staff members were bilingual in English and Spanish and approached dyads to determine eligibility and describe the study. Dyads were eligible if the adolescent was aged 11 to 14, was either Hispanic or African American, and identified as heterosexual or indicated sexual attraction to opposite sex partners. “Mothers” refers to both biological and nonbiological primary adult female caregivers. Once eligibility was established, dyads were informed about the opportunity to participate in a research project focused on improving parent-adolescent communication. Interested dyads signed informed consent forms for mothers and assent forms for adolescents. The research was approved by the New York University Institutional Review Board.

The FTT triadic intervention is adolescent centered, parent involved, and HCP endorsed. FTT was designed to reduce adolescent sexual risk behaviors and shape adolescent SRH through collective efforts of HCPs, provider extenders, and parents. Dyads randomly assigned to the experimental group received the standard of care plus the FTT intervention for mothers. Provider extenders (paraprofessionals trained in the FTT program) delivered the intervention for 45 to 60 minutes while the adolescent completed the wellness visit. The FTT intervention included 4 components: (1) a face-to-face session delivered by a provider extender to familiarize mothers with the intervention materials, emphasizing specific strategies for parent-adolescent communication about sex, condom use, and strategies for parental monitoring; (2) written intervention materials (a family workbook) for mothers to use with the adolescent at home designed to support parental mastery of effective communication about sex and monitoring or supervision strategies; (3) an HCP endorsement of the intervention to adolescents and mothers, conveying the importance of parent-adolescent communication regarding intervention content; and (4) 1 phone booster session 1 month after the face-to-face intervention in which provider extenders reinforced FTT messages. At no time did the intervention interfere with confidential time between the provider and adolescent patient.

The active control condition was similar to the experimental condition, except the content pertained to nutrition information instead of FTT. Mothers randomly assigned to the passive control group returned to the waiting room after completing the baseline survey. HCPs interacted with mothers and adolescents as they would with any patient not enrolled in the study.

Adolescents and their mothers were assessed at baseline and at 3 and 12 months post baseline via self-report surveys. After consenting and meeting with the HCP for their appointment, adolescents completed baseline questionnaires in a confidential room. Surveys included a measure of social desirability response tendencies and warm-up questions to familiarize participants with the measurement scales.³⁸  English and Spanish surveys were available to participants and were administered on the basis of language preference. Before the study, surveys were pilot tested for readability and comprehension.

Three months post baseline, project staff set up appointments for follow-up assessments in participants’ homes or at a mutually agreed on location in the community. Staff administered 3-month follow-up assessments to detect group differences in mother-adolescent communication about intervention content. Twelve months post baseline, a follow-up assessment was scheduled to determine group differences in adolescents’ reports of primary and secondary outcomes. For each assessment, mothers and adolescents received incentives of $15 and $10, respectively.

The primary outcome measures for this study were ever having had vaginal intercourse, sexual debut (defined as the first act of vaginal intercourse) within the past year, and condom use at last vaginal intercourse. At the 12-month follow-up, condom use at last sexual intercourse was assessed among sexually active adolescents.

The secondary outcome measures were related to adolescent reports of maternal monitoring and communication about sex. Maternal communication about sex was assessed by using a 5-point Likert-type scale. Adolescents reported their level of agreement with 3 items regarding communication within the previous 3 months about delaying sexual intercourse, STIs, and contraception use (including condoms). Maternal monitoring of knowledge was assessed by using a 5-point Likert-type scale via the item “My mother knows where I am after school.” Maternal monitoring of behavior was assessed with 4 statements regarding discussing plans with mothers, discussing when rules are broken, checking in on adolescents’ lives, and expectations for adolescents to call home when running late.

To assess the extent to which adolescents in the experimental arm were exposed to specific intervention content and to evaluate risk for potential contamination in the control arm, mean levels of exposure to specific FTT intervention content were measured at 3 months post baseline by using a 5-point Likert-type scale (described in Table 1).

A computer program was used to generate a randomly permutated scheme to allocate subject identification numbers to the experimental, passive control, and active control conditions in a 4:1:1 ratio. The principal investigator for the study generated the allocation sequence. After mothers completed the baseline assessment, a sealed envelope was opened by the project staff member who administered the assessment to assign dyads an identification number. Follow-up data collectors were blinded to participant conditions.

Formal evaluation of the RCT relied on linear and logistic regression for continuous and dichotomous outcome variables, respectively. Primary and secondary outcomes were analyzed by using intention-to-treat analyses. For intention to treat, all dyads randomly assigned at baseline were included in the analysis.³⁹  During preliminary analysis, no statistically significant differences in primary and secondary outcomes emerged between the active and passive control conditions; therefore, both control conditions were pooled into one control condition during subsequent analyses (Fig 1). Power analyses were based on pilot data.³³  For single degree-of-freedom contrasts of means, a sample size of ∼75 per group at α = .05 yielded power of 0.80 to detect a medium effect size, as defined by Cohen (standardized d value = 0.50).⁴⁰  For comparing percentages, a sample size of ∼280 per group yielded power of 0.80 for detecting a seven-percentage-point difference between 2 groups. To have sufficient power for subanalyses among sexually active youth, a sample size of 900 dyads was obtained for this study.

To address missing data, we employed multiple imputation with the mice package of the R statistical computing environment.^41,42  A total of 50 imputed data sets were generated, with primary outcome logistic regression analyses repeated for each imputation. Results were then pooled across imputations, with inferences and confidence intervals (CIs) considering within- and between-imputation variance. Diagnostic plots revealed good convergence of the data-augmentation algorithm for multiple imputations.

From July 19, 2010, to May 31, 2016, we screened 1256 mother-adolescent dyads for eligibility, of whom 984 were eligible for study participation. Eighty-four dyads declined to participate. Six hundred dyads were randomly assigned to the FTT group, whereas 300 dyads were randomly assigned to either the passive standard-of-care or active nutrition intervention control groups. Participants were followed for 12 months.

Of the 900 dyads participating in the study, 73.4% identified as Hispanic, 20.4% identified as African American, and 6.2% identified as mixed race (both Hispanic and African American). The mean (SD) age of mothers (range 18–78 years) was 38.8 (8.4) years, and the mean (SD) grade of adolescents (range grades 6–8) was 6.9 (0.9). Three percent of adolescents reported having had sexual intercourse. Complete demographic information for participating dyads, stratified by condition, is displayed in Table 2. Attrition was low; 91.4% of dyads interviewed at baseline completed the 12-month follow-up survey.

At the 12-month follow-up, 5.2% of adolescents who received the FTT intervention had engaged in sexual activity, compared with 18.0% of adolescents in the control group (relative risk [RR] 3.50; [95% CI: 2.30–5.32]; P < .05). Additionally, only 4.7% of adolescents who received the FTT reported sexual debut within the past 12 months, compared with 14.7% of adolescents in the control group (RR 3.12 [95% CI: 1.98–4.91]; P < .05).

At the 12-month follow-up, among adolescents who reported ever having sexual intercourse, 74.2% of adolescents who received the FTT intervention reported using a condom at the last sexual intercourse, compared with 49.1% of adolescents in the control group (RR 1.48 [95% CI: 1.05–2.08]; P < .05) (see Table 3).

At the 12-month follow-up, adolescents in the experimental group reported higher levels of parent-adolescent communication about delay of sex (unstandardized beta coefficient [B] = 0.64 [95% CI: 0.42–0.86]; P < .05), STIs and HIV (B = 0.60 [95% CI: 0.37–0.83]; P < .05), and using contraception and/or condoms (B = 0.37 [95% CI: 0.13–0.61]; P < .05) compared with adolescents in the control group. Furthermore, adolescents in the experimental group reported increased maternal monitoring knowledge and behavior, including monitoring of whereabouts after school (B = 0.15 [95% CI: 0.03–0.61]; P < .05), monitoring of plans with friends (B = 0.32 [95% CI: 0.15–0.50]; P < .05), discussion after breaking a rule or previous agreement (B = 0.28 [95% CI: 0.15–0.40]; P < .05), checking in regularly (B = 0.23 [95% CI: 0.05–0.40]; P < .05), and being expected to call when late (B = 0.13 [95% CI: 0.01–0.25]; P < .05) (see Table 4).

In the analysis of the intervention dosage, we explored the extent to which adolescents were exposed to specific FTT intervention content. Adolescents in the experimental group reported higher levels of mother-adolescent communication across topics (Table 1). The control arm of the study was evaluated for potential intervention content contamination. Less than 5% of the control arm reported previous exposure to FTT material.

We examined the potential moderating effects of race and/or ethnicity and sex on the overall treatment effects for each primary outcome. The treatment effects did not vary significantly for boys compared with girls or by race and ethnicity. To assess the generalizability of results across traditional and robust methods of analysis, we repeated all analyses with bootstrapping. No differences were detected. Furthermore, we conducted outlier and specification-error diagnostics for all analyses. The statistical checks returned no problematic results. Missing data were minimal (91% of the subjects randomly assigned at baseline completed the study). No statistically significant differential attrition between arms was observed.

With this study, we provide evidence for the efficacy of a clinic-based triadic intervention implemented with HCPs, provider extenders, and adolescents and their mothers for delaying sexual activity and reducing sexual risk behavior. Adolescents in the intervention group were significantly more likely to delay sexual debut, use a condom at the last sexual intercourse, and report increased maternal communication about delay of sex and maternal monitoring.

Noteworthy in our data was the enhanced condom use in the FTT group. Condoms represent an important multipurpose technology for the prevention of unintended pregnancy, STIs, and HIV.⁴³  However, condom use among ninth-graders has declined significantly over the past decade, with only 54.5% reporting condom use at the last sexual intercourse in 2017.⁴⁴  At the 12-month follow-up, only 49.1% of sexually active adolescents in the control group indicated condom use at the last sexual intercourse, similar to baseline rates. In contrast, adolescents in the experimental group reported significantly elevated rates of condom use at the last sexual intercourse (74.2%). With the current study, we provide evidence that triadic interventions are effective in shaping both adolescent condom use and parent-adolescent communication about contraception and/or condoms. Therefore, triadic interventions represent a novel mechanism for supporting national STI and HIV prevention efforts in the United States.

Taken as a whole, brief HCP endorsement of FTT content to mothers and adolescents represents a powerful, novel, and practical tool for promoting adolescent SRH. Previous research has revealed that parents are influential in adolescent SRH.^19–22,28  However, clinical responsibilities and associated time constraints limit HCPs’ ability to provide extensive consultations regarding parental involvement in adolescent SRH. Our results support the efficacy and added benefit of brief HCP endorsements of a parent-based intervention delivered by provider extenders. Provider extenders include medical assistants, social workers, counselors, etc, who can deliver the most time-intensive interventions (ie, face-to-face sessions and phone booster calls) and are easily embedded in the clinical setting. Given clinical realities, a provider extender–delivered SRH triadic intervention that includes a brief HCP component is a practical solution for addressing calls from parents and national health organizations for more parental involvement in adolescent SRH care. In summary, the FTT intervention for clinical settings represents an innovative, feasible, and low-cost tool⁴⁵  to strengthen partnerships between HCPs, parents, and adolescents to reduce negative SRH outcomes among younger adolescents.

Despite the strengths of this study, there are some limitations to our findings. First, the outcome measures relied on self-reports, which may be biased on the basis of social desirability. However, the random assignment of participants and the use of a social desirability index controlled for potential self-report bias. Second, we recruited participants in a single geographic and clinic setting, so generalizability to other settings may be limited. Furthermore, the follow-up (12 months) for the study was relatively short given that the mean age at sexual debut is 17 years.⁴⁶  Findings must be interpreted with caution when generalizing to older adolescents. Third, the research was limited to mother-adolescent dyads, as mothers were most likely to be present at their adolescent children’s health care clinic appointments. Future researchers should explore strategies to increase paternal involvement in adolescent SRH and ideal settings to recruit fathers.⁴⁷  Fourth, for subanalyses of condom use among sexually active youth, we were only able to include a proportion of the 900 enrolled dyads. Future researchers should extend the findings of this study to understand how triadic interventions can impact condom use among the more sexually active older age groups. Fifth, future research should specifically be focused on sexual minority youth and their families. Lastly, primary outcomes were limited to adolescent reports of vaginal intercourse. Future researchers should explore how triadic clinic-based interventions influence adolescent oral and anal sexual behavior.

The FTT triadic intervention implemented with HCPs, mothers, and adolescents delayed sexual debut and reduced sexual risk behaviors among Hispanic and African American adolescents. Intervention delivery by provider extenders, combined with a brief HCP component, represents an innovative, brief, and comparatively low-cost option⁴⁵  for broad dissemination of triadic SRH interventions in pediatric and adolescent health clinical settings. Broad uptake and implementation of the FTT triadic intervention in pediatric and adolescent health clinics represents a potential opportunity to achieve progress toward national goals to reduce STIs, HIV, and unplanned pregnancies among adolescents across the United States.^7–9 

B

unstandardized beta coefficient

CI

confidence interval

FTT

Families Talking Together

HCP

health care provider

RCT

randomized clinical trial

RR

relative risk

SRH

sexual and reproductive health

STI

sexually transmitted infection