Even with the ongoing decline in U.S. teenage pregnancies over several decades, some 2,000 teenagers become pregnant every day.¹ In addition, an estimated 40% of sexually active adolescent females have an STD.¹ Long-standing disparities in teenage pregnancy and STD rates still exist by race and ethnicity.^{2, 3} Educational programs that target parents are an alternative or an enhancement to school-based adolescent sexual health education.⁴ Parent-based interventions aim to reduce adolescent sexual risk behaviors (e.g., early sexual debut, nonuse of condoms) by bolstering parental protective factors such as parent–child communication about sexual health⁵ and parental monitoring,⁶ as suggested by the analytic frameworks of Burrus et al.⁷ and Hutchinson and Wood.⁸ For example, one study found that an increased level of sexual risk communication between mothers and daughters (12–19 years old) was associated with reduced levels of sexual intercourse and unprotected sex.⁹ Although findings from some U.S. intervention trials suggest that improvements in parent–child communication can influence adolescent sexual behaviors,^10-14 the empirical evidence to support a causal claim is lacking.

Past reviews of parent-based adolescent sexual health trials have used widely different inclusion criteria and search strategies,^15-19 which led to large variations in both the U.S. and the non-U.S. studies examined; as a result, only three trials were included in all reviews. Moreover, these reviews based their conclusions on descriptive findings and the percentages of studies with statistically significant outcomes, and did not include a meta-analysis. New studies have been published since these reviews were completed.

It is not surprising, then, that prior reviews reached different conclusions, some declaring that findings are positive^15-17 and others that findings are mixed.^{18, 19} In summarizing 15 trials, Atienzo et al.¹⁵ described them as showing significant increases in parent–child communication; parental monitoring and supervision; adolescents’ perceptions of rules and of family support; and intentions to postpone intercourse and use contraceptives and condoms. These authors found a reduction in adolescents’ self-reported sexual relations associated with parent-based interventions. Wight and Fullerton¹⁶ found that interventions were associated with improvements in parent–child interaction and adolescents’ sexual knowledge and attitudes; about half of the 44 studies they reviewed reported improvements in adolescent sexual behavior outcomes. Bastien et al.,¹⁷ reviewing 23 Africa-based trials, reported positive findings, largely increases in frequency of and parental comfort with sexual health discussions. Akers et al.¹⁹ found that among 12 trials, intervention parents showed improved communication, as measured by frequency, quality, intentions, comfort and self-efficacy for communicating; however, no improvements in parental attitudes toward communicating were seen. The Downing et al.¹⁸ review of 17 studies found no improvements in parental attitudes toward communication, and reported inconsistent association with adolescent sexual risk behaviors; no association was found between increased parent–child communication and a decline in sexual risk behaviors.

In addition to the need for stronger analytic methods in systematic reviews, it is important to describe the theoretical foundation on which interventions are based. The importance of theory in intervention development and implementation is well established across behavior change interventions in general,^{20, 21} in adolescent sexual health interventions^{7, 8, 22} and in parent-based interventions.^{7, 8} Theory is used to explain personal and environmental determinants of behavior related to a specified health problem, create a causal model for bringing about positive outcomes, select intervention methods and practical delivery applications to achieve behavior change, evaluate interventions and describe interventions’ critical components.^{23, 24} For example, the theory of planned behavior has been used to understand adolescent and parent behaviors,²⁵ and social cognitive theory has been used to inform intervention components, methods and applications.²⁶

Previous reviews have not adequately described the use of theory in intervention development (e.g., how theory informs determinants, methods and applications) or interventions’ doses or delivery mechanisms, in part because published trials do not consistently report methods used to change behavior.²⁷ An examination of the theoretical underpinnings of behavioral determinants, behavior change methods, the application of methods, intervention setting and delivery mode would allow for a more comprehensive description of the critical components of parent-based adolescent sexual health interventions.²⁸ Moreover, a better understanding of these components would provide insight into how personal and environmental factors that influence behavior can be changed²⁹ and how methods (e.g., modeling) and their applications (e.g., a video demonstrating effective parent–child communication) can be successfully implemented.

This review addresses the following research questions: First, what is the range of settings, target populations, delivery modes and doses in parent-based adolescent sexual health interventions? Second, what theories and models have been used to inform the development of such interventions, and to what extent have they been used? Third, what is the range of determinants and behavioral outcomes addressed by these interventions? Fourth, what is the range of theory-based methods and applications that have been used? Finally, what is the impact of such interventions on parent–child communication and adolescent sexual behavior?

METHODS

Systematic Review

Trials eligible for inclusion in this review reported findings from parent-based adolescent sexual health interventions, were written in English and were published in peer-reviewed journals between January 1998 and July 2013. We used Kirby and Miller's 2002 descriptive systematic review of this type of intervention as a guide.³⁰ We included all adolescent sexual health intervention trials that targeted parents, and did not limit eligibility to those that focused only on parent–child communication. Since the cultural climate for this topic is vastly different in the United States than in Europe, Africa, Australia and Latin America, we limited the review to U.S. trials.

The search strategy was developed with the assistance of a public health librarian with expertise in conducting and documenting searches for systematic reviews. Databases searched (and interfaces) were Medline (Ovid), PubMed (NLM), PsycINFO (Ovid), CINAHL (EBSCOhost) and ERIC (EBSCOhost). Concepts included in the search were "parent," "adolescent," "sexual behavior" and "evaluation." The search strategy for each database, which differed to account for specialized index terms, was reviewed by the authors and the librarian, and was archived for future inquiries and updates. An exhaustive list is available upon request. We also searched the reference lists of included trials for additional intervention studies and articles related to the development and theoretical underpinnings of the interventions.

Retrieved citations were screened, duplicates were eliminated and the remaining citations were organized in a database (RefWORKS). The first author reviewed titles and abstracts and applied the eligibility criteria; full-text articles were reviewed to determine final eligibility. A random sample of 50 titles and abstracts was independently screened by a second reviewer to assess the reliability of the selection procedure.

Data abstracted from each study were study design, sample size, recruitment methods, setting, intervention aims by intended audience, components, delivery mode, dose (number of sessions and duration), behavior change methods and practical applications, theory and measured outcomes. We evaluated the extent of theory use by employing an adaptation of the Michie et al. coding scheme.²⁷ Trials were assigned a level from 0 (theory not mentioned at all) to 4 (theory use described in detail relative to intervention, determinants of behavior and theory-relevant constructs). Theoretical methods and applications were abstracted and coded using the Bartholomew et al. taxonomy.²⁹ This study was approved by the institutional review board of the University of Texas Health Science Center at Houston.

Meta-Analyses

Meta-analyses were performed on subsets of trials from the systematic review that had a contemporaneous control group and conceptual similarity in the outcomes of interest: parent–child sexual health communication and parental comfort with sexual health communication. (Adolescent-level measures—e.g., sexual intercourse, condom use—were too diverse to be included in a meta-analysis.) For these trials, we abstracted the study design, sample size and measurement properties of outcomes (i.e., indicators and types of scales used). To assess estimated effect sizes for the individual trials, we used Cohen's d to calculate standardized mean differences (e.g., the standardized difference in posttest mean scores of parent–child sexual health communication between the intervention and control groups divided by the pooled standard deviation).³¹ When trials had multiple follow-up points, we used the most proximal follow-up measures beyond immediate postintervention. For trials with more than one intervention group and a single control or comparison group, we selected the parent-only intervention that was theoretically most effective, usually indicated by the authors.

RESULTS

Trial Characteristics

A total of 28 parent-based adolescent sexual health intervention trials, reported in 62 articles, met the eligibility criteria (Table 1).^{10-14, 32-54} Interventions most frequently targeted parents of adolescents younger than 16 years old, 18 enrolled mostly minority parents, 20 took place in the community and eight provided self-paced activities at home.

Interventions used a variety of delivery modes and were most frequently offered over multiple face-to-face sessions. Eighteen trials employed multiple group sessions—either sessions for parents only, separate parent and child sessions, sessions for parents and children together, or a combination. Eight interventions used mixed delivery modes (brief individual sessions, self-paced materials and media exposure). The dose ranged from 30 minutes to 24 contact hours, and averaged about seven hours.

Of the 23 theory-based trials, 20 reported extensive use of theory (categorized as level 4), including 12 that employed social cognitive theory, the most frequently used.²⁶ Twenty theoretical models and frameworks were reported as underlying intervention development and dissemination. Theories informed multiple aspects of the interventions by providing a framework for targeting parents as a potential influence on adolescent behavior (e.g., ecodevelopmental theory);⁵⁵ constructs that influence parental behaviors (e.g., social cognitive theory); and delivery mode, setting, methods, applications and dissemination (e.g., diffusion of innovation). Planned behavior⁵⁶ and reasoned action⁵⁷ theories were used in seven trials.^{13, 32, 36, 40, 42, 44, 47} Thirteen trials reported using constructs from more than one theory, model or framework,^{11, 13, 32, 34, 36-38, 40, 42-45, 47} and one trial reported using five theories.¹³

Interventions employed a variety of methods to meet objectives: Nineteen used modeling and discussion, 16 used guided practice, seven used verbal persuasion and five used message tailoring. These methods were delivered via face-to-face interactions, videos, CDs and homework. Most theory-driven methods were based in social cognitive theory (e.g., modeling, guided practice, enactment, verbal persuasion, and physiological and affective change).²⁹ Detailed explanations of how the theoretical models were operationalized in intervention activities and of the process used to select theory, methods and applications were rarely given in the publications.

Determinants and Behavioral Outcomes

Interventions targeted several parent-level behavioral determinants to increase parent–child sexual health communication. Fifteen interventions targeted parental self-efficacy for communication, 11 targeted communication skills and attitudes toward communication, and 10 focused on positive outcome expectations. Interventions focused on other parental practices as well: increasing monitoring (14), facilitating setting and enforcing family rules (11), increasing parental involvement and support (10), and mobilizing social networks (five). Some interventions aimed at preparing parents to teach effective refusal and negotiation skills (six); help youth improve decision making, self-regulation and self-control, and increase their awareness of parental values and behavioral expectations (five); and help them improve their self-efficacy, social problem solving and coping with peer pressure (four). Authors often did not report the methods or applications used to improve parents’ ability to address these key adolescent behavioral determinants.

All 28 interventions aimed to improve effective parent–child communication; however, they addressed different domains of communication, targeted different behavioral determinants and used a variety of methods and applications. Each intervention reported a positive effect on at least one parent or adolescent outcome. Fifteen interventions reported a positive impact on parent–child communication frequency, and 11 on parental comfort with communication about sex, the most frequently reported parent-level outcome. Of the interventions that measured adolescent outcomes, five reported positive outcomes for reducing sexual risk (e.g., decreased sexual activity), and five for increasing condom use.

Results of Meta-Analyses

Eleven trials were selected for meta-analysis of parent–child sexual health communication, and nine for meta-analysis of parental comfort with sexual health communication (Table 1). Seven trials were included in both meta-analyses. All included interventions were conducted either face-to-face or through self-paced activities.

To arrive at a single, independent outcome measure for each trial, a hierarchical approach was implemented that favored adolescent-reported measures, which were considered more relevant to sexual health behavior than parent-reported measures. For example, Evans et al.³⁵ found that only 33% of sons aged 10–14 agreed with their mothers’ reports of mother-son sexual communication. When adolescent-reported measures were not available, parent-reported measures were used, a strategy that has been employed in previous meta-analyses.⁵⁸

Cohen's d⁵⁹ indicated that the intervention groups were significantly more likely than the control or comparison groups to report positive effects. However, there was a range (0.1–1.7) in effect sizes for parent–child sexual health communication (Figure 1). The Cohen's d for overall intervention impact on this outcome was 0.5, signifying a medium effect size.⁶⁰ The range of effect sizes for parental comfort with sexual health communication was even wider (0.01–2.1), and the overall effect was large (0.7—Figure 2).⁶⁰ These overall impacts mean that an intervention group parent was 68% more likely than a control group parent to have an increased communication score, and was 75% more likely to report increased comfort with communication.⁶¹ The data suggest a positive direction of effect across all intervention delivery modes—short, self-paced, short group and long group—and show that self-paced, lower dose interventions were not consistently less effective than higher dose interventions.

Because of the inconsistencies in outcome measures, dose, delivery modes and setting, tests for heterogeneity between studies indicated, as expected, a high level of inconsistency when we evaluated the same effect across trials (I², 94% for parent–child sexual health communication and 98% for parental comfort with communication). To address the high level of heterogeneity, we used a random effects model to weight the results of the individual studies before calculating the summary measure.

DISCUSSION

Setting and Target Population

The parent-based adolescent sexual health programs examined in this review most frequently occurred in a community setting with parents of minority youth. Our findings point to several gaps in the range of published parent-based trials. We found no trials that targeted mothers and sons, although 24% of youth live in single-mother households.⁶² Nor did we identify trials targeting fathers and daughters, despite evidence that paternal communication about sex influences adolescent sexual behaviors separately from maternal communication,⁶³ and has been suggested to be qualitatively different.⁶⁴ Moreover, fathers are also in need of education and skill building in sexual health communication.⁶⁵ In addition, no trials were found that were developed for faith-based organizations or services, despite evidence that faith-based programs can improve health outcomes and program reach.⁶⁶ Finally, no interventions were identified that targeted parents of lesbian, gay, bisexual or transgender youth or of youth living with custodial grandparents, both of which are high-risk adolescent groups.^67-69

Dose and Delivery Mode

Most of the included interventions were resource- and dose-heavy. This poses several concerns related to the challenges of disseminating resource-intensive programs and expanding their reach. First, intensive face-to-face group sessions require substantial funding, extensive procedural manuals, training of facilitators to assure program fidelity, community-level planning and laborious coordination. Second, interventions that require attendance at multiple sessions often face recruitment and retention challenges. For instance, at-risk populations may have a high level of life chaos and feel that the intervention is irrelevant or unhelpful in addressing their prioritized needs.⁷⁰

Indeed, the meta-analyses indicate that the effects of the interventions using self-paced, lower dose, easily disseminated modes of delivery (e.g., computer-based) were not consistently smaller than those of high-dose interventions. This provides some support, albeit tenuous, for the development and testing of a range of low-dose intervention delivery modes that could increase reach to the parents of the highest risk groups by reducing barriers associated with face-to-face modalities. Our systematic review found that most parent-based interventions relied on group-based, multisession delivery modes. One promising way to increase reach is through the use of technology. Computer-based programs can reduce common barriers, such as work and family obligations, time and transportation. Use of the Internet bypasses access barriers, is familiar and comfortable to young parents, allows for individualized and tailored messages, engages active learning and is accessible even to low-income households.^{71, 72} While there are challenges to disseminating high-dose interventions, further research is needed to determine if low-dose ones and methods other than face-to-face approaches can lead to similar positive outcomes and whether those outcomes can be sustained over time.

Use of Theory

Although most interventions used theory extensively to develop their scopes and aims, others did not indicate any use of theory in development, implementation or dissemination. Still other interventions employed theory to target a certain audience by race or ethnicity (e.g., community-as-mother theory) or by gender (e.g., gender theory). Theory was used to determine which constructs to target and to understand behavioral pathways.²⁰ We found that well-constructed interventions that reported high theory use most frequently employed constructs from more than one theory or framework. Several trials cited theory to support targeting their intervention to parents as a means to change adolescent sexual health behaviors and outcomes, yet they did not explicitly specify theoretical methods and applications. Some trials described using the theory of planned behavior. However, a previous systematic review of the use of this theory found that it was employed most frequently to understand behavior, but not to design intervention components, methods or applications.²⁵

While there are systematic ways to use theory to develop interventions, few parent-based youth sexual health intervention trials explicitly described this process in their publications, and few trials appear to use theory to determine appropriate methods, practical applications or delivery modes of the interventions.²⁹ Another study shows that social cognitive theory and self-regulation theory are commonly used to inform behavior change methods (e.g., modeling, persuasion, self-monitoring).⁷³ However, using social cognitive theory may not increase intervention effectiveness,⁷⁴ and frequently trials did not report the application of the theoretical methods used.

Most interventions aimed to increase effective parent–child sexual health communication. Yet other parental constructs, such as parental attitudes, parent–child relationship quality and monitoring, have been correlated with adolescent sexual behaviors. For instance, in other research, youth who perceived parental permissiveness were more likely to engage in sexual behaviors, whereas perception of parental disapproval reduced the level of sexual risk behaviors.^75-77 Additionally, parental nurturing and supportiveness have been associated with delay of sexual debut and condom and contraceptive use,^{78, 79} and effective parental monitoring has been found to reduce situational opportunity.^{80, 81} These examples highlight the importance of addressing multiple protective parenting practices when developing parent-based adolescent sexual health interventions. While the current study provides evidence that an intervention can improve parent–child sexual health communication, our findings do not allow us to conclude that improving communication directly affects adolescent sexual behaviors or outcomes.

The trials frequently did not justify their use of theory, their choice of theory⁸² or the process employed to determine the most appropriate theory, methods and applications. Despite the proliferation of parent-based adolescent sexual health interventions and related systematic reviews, this review highlights important gaps in our understanding of the range of intervention characteristics and effects on parent and adolescent behavior. By using the Michie et al. theory coding scheme,²⁷ we were able to quantify the extent of theory use. In using the Bartholomew et al. taxonomy,²⁹ we were able to assess the methods and applications used to implement the interventions. Many interventions characterized by high theory use described using methods we would expect to see. For instance, interventions used modeling and guided practice operationalized through applications such as video demonstrations and role-playing to improve skills, and persuasion and discussion operationalized through applications such as persuasive messages and small group discussion to change beliefs about and attitudes toward communication.

To improve parent-based adolescent sexual health programming, a better understanding of theory use and how to determine intervention effects is necessary. It is critical that we identify effective methods and practical applications of interventions when developing and evaluating parent-based initiatives. While understanding the theoretical constructs and methods of an intervention is helpful, reporting the parameters and applications of the methods is an essential next step,²⁹ albeit beyond the scope of this study. The use of existing taxonomies to select methods and applications that operationalize theory may strengthen an intervention.

Meta-Analyses

Consistent with a previous descriptive systematic review,¹⁹ our meta-analyses revealed that intervention group parents and their adolescents reported significant increases in the amount of and comfort with parent–child sexual health communication in most of the primary studies. However, effect sizes ranged widely across trials. All but one trial⁵³ with a large effect scored 4 for theory use (highest). Further analysis is needed to determine the extent to which the size of the effect differs depending on theory use, methods, applications and dose. We identified some potential sources of heterogeneity, including inconsistencies in outcome measures, dose, intervention modalities and setting. This may be more important in conducting meta-analyses of community-based, parent-based adolescent sexual health intervention trials than the heterogeneity of the findings themselves.⁸³ Developing consensus on the optimal outcome measures would reduce inconsistencies in evaluating effects and allow for further analysis across trials. Nonetheless, the estimates for communication and comfort suggested the same positive direction of effect across all trials, regardless of delivery mode. Indeed, compared with control parents, intervention parents had an elevated likelihood of reporting increased communication and comfort with communication.

While some of the trials measured adolescent-level outcomes, too few consistently measured outcomes to allow us to pool them and assess their effects on adolescent behaviors. Additional trials that measure adolescent behavioral outcomes are needed before it is possible to assess the impact of parent-based intervention effects on adolescent-level outcomes. The data suggest a positive direction of effects for parent-level outcomes, and these were not consistently smaller for low-dose than for higher dose interventions. However, the evidence of positive effects is not as strong for short, self-paced interventions as for long, group-based ones. Further research is needed to determine whether low-dose interventions or delivery modalities other than the traditional group, multisession format result in positive, sustained effects.

Limitations

The high levels of heterogeneity observed in the meta-analyses suggest caution when interpreting the findings. We believe this heterogeneity is related to wide variations in participant type, trial design, sample size, mode of intervention delivery and communication measures used; we compensated for it by using a random effects model and by subdividing trials into groups with similar outcomes.

Further, there is a risk of publication bias since negative trial findings are less likely to be submitted and published, and thus were not included in this study. To reduce study limitations related to inclusion bias, our systematic search strategy was conducted in collaboration with a public health librarian. Use of standardized intervention descriptions, as outlined by Bartholomew et al.²⁹ and Michie et al.,²⁷ facilitated the consistent characterization of theory use, methods and applications. We observed general patterns of theory use, which may support further exploration of such use in intervention development. For example, five of the six trials with the largest effect sizes in both meta-analyses scored highest in theory use. Although this aspect was central to the descriptive systematic review, and the extent of theory use was categorized, it was not included in the meta-analyses. In addition, searching only English-language journals may have led to the exclusion of pertinent trials published in other languages. By restricting selection to peer-reviewed publications to assure reporting quality, we may have missed relevant gray literature, particularly that which might have reported negative findings. Finally, because of the sensitive nature of the topic, recruiting an adequate sample size may be a general challenge for sexual health researchers.

Conclusions

Our findings provide preliminary support for the effectiveness of parent-based interventions on parent–child sexual health communication, and point to future research needs.⁷ Lower dose, barrier-reducing interventions, such as self-paced ones, hold promise for the dissemination of parent-based interventions, have been reported to be acceptable to minority families³² and warrant further exploration. More studies with longer follow-up periods that measure adolescent risk behaviors and sexual health outcomes, such as cases of STDs and pregnancy, are needed, and future trials should provide justification for the use of a particular intervention development theory. Greater analysis of the use of theory to understand barriers and design interventions, to explore mediating pathways and moderators, and to examine the impact of theory use on adolescent and parent outcomes would advance the science of implementation research.^{24, 82} Developers and researchers who use theory to design interventions should report, using a standard taxonomy, the theoretical constructs and methods employed to promote change,^{23, 29} and should include the practical applications of those methods to populations and settings. Furthermore, a study of all children in a household could help determine if programs that target parents have the potential for exponential effects on multiple children. Finally, a cost analysis of parent-based interventions would be useful to help determine the potential reach of effective programs.

Acknowledgments

The authors thank Helena VonVille for her valuable contributions to the development and documentation of the search and citation management and selection, and Melissa Peskin for cross-checking trial inclusion screening. Shirley Bluethmann was supported by the Susan G. Komen Foundation (grant KG111378) and the Cancer Education and Career Development Program at the School of Public Health, University of Texas Health Science Center at Houston, funded by the National Cancer Institute (grant R25T 2R25CA57712). The findings and conclusions of this article are those of the authors and do not necessarily represent the official positions of the funding organizations.