Screening and inclusion and exclusion criteria for studies in the meta-analysis.
Each study is followed by 2 letters: the first represents the sex of the sample (B, all boys; G, all girls; and M, mixed), and the second represents the race/ethnicity of the sample (B, black; L, Latino; M, mixed; and W, white). The diamond indicates the overall weighted mean effect across all studies (r = 0.10; P < .001).
aSample from Uganda.
bSample from Malawi.
cSample from Ghana.
dSample from Burkina Faso.
eTable. Descriptive Characteristics of 52 Studies and 71 Independent Effects Included in the Meta-Analysis
Widman L, Choukas-Bradley S, Noar SM, Nesi J, Garrett K. Parent-Adolescent Sexual Communication and Adolescent Safer Sex BehaviorA Meta-Analysis. JAMA Pediatr. 2016;170(1):52-61. doi:10.1001/jamapediatrics.2015.2731
Parent-adolescent sexual communication has received considerable attention as a factor that can positively affect safer sex behavior among youth; however, the evidence linking such communication to youth contraceptive and condom use has not been empirically synthesized.
To examine the effect of parent-adolescent sexual communication on safer sex behavior among youth and explore potential moderators of this association.
A systematic search of studies published from database inception through June 30, 2014, using the MEDLINE, PsycINFO, and Communication & Mass Media Complete databases and relevant review articles yielded 5098 studies, of which 52 studies with 25 314 adolescents met the study eligibility criteria. Analysis was conducted from July 1, 2014, to July 27, 2015.
Studies were included if they sampled adolescents (mean sample age ≤18 years), included an adolescent report of sexual communication with one or both parents, measured safer sex behavior, and were published in English.
Data Extraction and Synthesis
Correlation coefficients (r) and 95% CIs were computed from studies and meta-analyzed using random-effects models.
Main Outcomes and Measures
Safer sex behavior, including use of contraceptives or condoms.
Fifty-two articles, including 71 independent effects representing more than 3 decades of research on 25 314 adolescents (weighted mean age, 15.2 years) were synthesized. Across studies, there was a significant weighted mean effect (r = 0.10; 95% CI, 0.08-0.13) linking parent-adolescent sexual communication with safer sex behavior, which was statistically heterogeneous (Q = 203.50, P < .001, I2 = 65.60). Moderation analyses revealed larger effects for communication with girls (r = 0.12) than boys (r = 0.04) and among youth who discussed sex with their mothers (r = 0.14) compared with their fathers (r = 0.03). Effects did not differ for contraceptive vs condom use or among longitudinal vs cross-sectional studies, indicating that parent sexual communication had a similar effect across study designs and outcomes. Several methodological issues were identified in the literature; future studies can improve on these issues by measuring parent-adolescent communication with robust, multi-item measures, clearly specifying the target parent, and applying multimethod longitudinal designs.
Conclusions and Relevance
Sexual communication with parents, particularly mothers, plays a small protective role in safer sex behavior among adolescents; this protective effect is more pronounced for girls than boys. We discuss the implications for practice and make suggestions for future research on parent-adolescent sexual communication.
Risky sexual behavior among US adolescents is a serious public health problem. Although adolescents make up only one-fourth of the population that is sexually active, they acquire half of all sexually transmitted infections (STIs).1 This number amounts to 9 million STIs, including more than 8300 new cases of human immunodeficiency virus (HIV) infections, each year.1 In addition, adolescents are at heightened risk of unintended pregnancy.1,2
Parent-adolescent sexual communication has received considerable attention as one factor that could positively affect safer sex behavior among adolescents, including the use of contraception and condoms. There are practical and theoretical reasons why parents may be agents of sexual socialization for young people. From a practical perspective, parents may play a critical role in conveying sexual information and can exert significant influence on adolescents’ sexual attitudes, values, and beliefs regarding risks.3,4 Parents may also provide a powerful model of open and honest communication about sexual health issues, which teens may emulate in their own sexual relationships.5
Parents’ influential role on child and adolescent behavior is also widely accepted in developmental and health behavior theory. Bronfenbrenner’s6 classic Ecological Systems Theory of human development suggests that individuals live within a series of nested systems—including the family system—that are dynamic, reciprocal, and can directly and indirectly influence behavior. Grounded in this approach, parent-adolescent sexual communication has increasingly been implicated in health behavior theories that explain sexual behavior among youth,7- 9 such as the multisystem perspective of sexual risk behavior among adolescents.9
Although practical and theoretical considerations suggest that parent communication should be strongly associated with safer sex behaviors among adolescents, there is surprising inconsistency in the empirical literature.3,9- 12 While several studies have found moderate, positive associations between parent communication and youth contraceptive or condom use,13- 16 other studies have found nonsignificant or even negative effects.17- 19 Furthermore, while it is possible that parental communication about sex can be protective for youth, open sexual communication often does not take place. Instead, embarrassment, inaccurate knowledge, or low self-efficacy may prevent some parents from engaging their children in honest and supportive conversations about sexual behavior.20 These barriers may explain why nearly one-fourth of youth report that they have not discussed sexual topics with a parent3,21- 23 and why even fewer have had meaningful, open conversations about the sexual issues that are critical to their long-term health.
The purpose of this meta-analysis was to synthesize the literature and determine the mean weighted association between parent-adolescent sexual communication and youth contraceptive and condom use. Findings from such an analysis are critical to the growing body of interventions that target adolescent-parent dyads to prevent HIV infections, STIs, and unintended pregnancies24- 26 and may be of considerable interest to researchers, educators, family practitioners, and parents themselves. To our knowledge, no such meta-analysis has been published to date despite several narrative reviews and calls for better synthesis of the literature.3,9- 12 A first goal of this meta-analysis was to estimate the magnitude of the association between parent communication and safer sex behavior among adolescents. We focused on safer sex behavior (ie, contraceptive and condom use) given the importance of these behaviors to the prevention of HIV infections, STIs, and unintended pregnancies.27
Given the heterogeneity in this literature, a second goal of this meta-analysis was to examine several potential moderators of the association between communication and safer sex behaviors. Two key moderators examined were sex of the adolescent and sex of the parent. Given existing evidence, we expected to find a more robust association between communication and safer sex behavior for girls compared with boys28- 31 and for communication with mothers compared with fathers.19,32 Several additional demographic and measurement moderators were also explored. These factors have been examined in prior work and are of direct relevance to family communication interventions. They included adolescent age, race/ethnicity, study location (United States vs non–United States), study design (cross-sectional vs longitudinal), communication measurement characteristics (source, topic, format, and number of items), and safer sex outcome (use of contraceptives or condoms).
This meta-analysis examined the link between parent-adolescent communication about sex and safer sex practices among youth.
More than 30 years of data were analyzed from 52 studies and 25 314 adolescents.
There was a significant mean effect linking parent-adolescent sexual communication with safer sex behavior among youth.
The association between parent communication and safer sex behavior was stronger for girls and for teens who discussed sexual topics with their mothers.
A detailed search was undertaken to locate relevant articles. First, comprehensive searches of the MEDLINE, PsycINFO, and Communication & Mass Media Complete databases were conducted from their inception through June 30, 2014, using the following combination of keywords: (adolescen* OR teen* OR youth OR middle school OR high school) and (communicat* OR discuss* OR negotiat* OR assert* OR talk OR influence) and (contracept* OR birth control OR condom* OR unprotected sex OR sex* risk OR safe* sex). Then, additional studies of potential relevance were located by examining review articles related to sexual communication.3,9- 11,33- 36 Analysis was conducted from July 1, 2014, to July 27, 2015.
Studies were included if they met the following criteria: they sampled adolescents, defined as a mean sample age of 18 years or younger, and no participants were older than 24 years37; included an adolescent report of sexual communication with one or both parents (studies that focused exclusively on parent-reported communication were excluded); measured safer sex behavior, including contraceptives (sometimes referred to as birth control), condoms, or unprotected sex; reported an association between parent-adolescent communication and safer sex behavior (when bivariate associations were not reported, authors were directly contacted for this information); and were published in English. We excluded articles that used a composite variable for sexual risk taking whereby it was not possible to determine the outcome of contraceptive or condom use (eg, combining number of sexual partners or abstinence along with condom use into a single composite variable).38,39 In addition, in a few instances, there were multiple relevant articles that used the same data set; in these cases, the article with the most complete data relevant to this meta-analysis was included.
The initial search produced 5098 scientific references. After a review of titles and abstracts, this sample was reduced to 510 articles. The full text of these 510 articles were then located and reviewed. After applying all selection criteria, the final sample consisted of 52 articles (Figure 1). Within this final sample, several articles reported results separately for independent samples, including 10 studies with analyses separated by sex, 2 studies separated by race/ethnicity, and 1 study separated by country (eTable in the Supplement). Independent effect sizes were calculated for each sample in these cases, resulting in 71 independent effect sizes.40
Most studies reported a single indicator of communication and safer sex behavior. When multiple indicators were reported, several steps were taken to avoid violating the assumption of independence that underlies the validity of meta-analyses.41 First, when studies reported contraceptive or condom use with a frequency score as well as use at first or last intercourse, we used the frequency variable to calculate an effect size because this variable is more representative of the overall pattern of contraceptive use.42 Next, when studies reported more than 1 measure of parent-adolescent sexual communication, we analyzed the data in 1 of 2 ways. For most analyses, we averaged these communication variables to maximize the use of available data and not favor one measure over another (ie, overall weighted effect size and comparisons by age, sex, ethnicity, and study location). However, for analyses that examined moderation by communication measurement characteristics, using this averaging approach would have resulted in a loss of specificity of variables and thus a loss of data. For these analyses, we used a random number generator to randomly select 1 variable for inclusion.37,41 The same procedure was used to handle the 3 studies that reported both general contraceptive use and condom use.17,43,44 Specifically, we averaged the effect of communication on these sexual health outcomes for primary analyses, but we used a random number generator to select 1 outcome variable from each study when we examined the type of outcome (contraceptive use vs condom use) as a moderator.37,41
Two of us (S.C.-B. and J.N.) independently coded the following data from each study: demographic and sample characteristics, sexual communication measurement characteristics (ie, communication topic, format, source, and number of items), and measurement of safer sex behavior (ie, type of safer sex behavior and measurement time frame). Communication topic was coded into 4 possible categories, including communication about contraceptive or condom use, pregnancy, STIs and/or HIV infection, or general sex-related topics (eg, discussing “sex”). Communication format was coded into 3 categories: behavioral frequency of communication (ie, ever or never, or indication of frequency of sexual communication), quality of communication (ie, perceived comfort, ease, or openness of communicating), and self-efficacy (ie, perceived confidence in ability to communicate about sexual behavior). Finally, communication source was coded based on the parent with whom the adolescent had discussed sex (mother, father, or one or both parents). Regarding safer sex behavior, the type of behavior was coded as general contraception, condom use, or unprotected sex (reverse coded to keep direction of effects consistent). The measurement of the time frame of safer sex behavior was coded as lifetime, past 6 months, past 3 months, first sexual intercourse, or last sexual intercourse. The mean percentage agreement between coders across all categories was 96%. Discrepancies between coders were resolved through discussion with the lead author (L.W.).
A correlation coefficient (r) was used as the indicator of effect size (range, –1.0 to 1.0).45 Effect sizes based on correlations can be interpreted as small (0.10), medium (0.25), or large (0.40).46 When bivariate correlation coefficients were reported in an article, they were directly extracted. If correlation coefficients were not reported, then appropriate formulas were used to convert other statistics (eg, t tests, summary statistics, and odds ratios) to approximate correlation coefficients.45,47 When none of the statistics could be converted to a correlation coefficient or when only multivariable analyses were reported, the study authors were contacted and appropriate raw data were requested. To keep effect sizes consistent and interpretable, values were transposed so that positive correlations always indicated a positive association between communication and safer sex behavior.
Once study characteristics were coded and effect sizes were extracted, a Fisher r to z transformation was performed.45 These values then were weighted by their inverse variance and combined. We used random effects meta-analytic procedures for the primary analysis across all 71 independent effect sizes; this procedure allowed for the possibility of differing variances across studies.41 After analyses were complete, the effect sizes were transformed back to correlation coefficientsfor presentation. The Q statistic and I2 were used to examine whether significant heterogeneity existed among effect sizes. Effect sizes for hypothesized moderators were calculated along with their 95% CIs, and those effect sizes were compared using the Qb statistic. For these analyses, mixed effects models were used to allow for the possibility of differing variances across subgroups. These models use random effects assumptions while stratifying the effect sizes by fixed factors, such as sex and study location.41 Analyses were conducted using Comprehensive Meta-Analysis software, version 2 (Biostat, Inc).
The eTable in the Supplement provides a summary of the 52 studies and 71 independent effect sizes represented in this meta-analysis, including sample characteristics and moderator variables. Across the studies, 25 314 participants were included (weighted mean age, 15.2 years). Of the 71 independent effect sizes, most (k = 52) were based on reports of communication with one or both parents; 19 studies specified whether communication was with a mother or father. Similarly, many studies (k = 47) asked about general sexual communication, with 24 studies assessing more specific topics, such as condom use, HIV infection and STIs, or pregnancy. In addition, more than half the studies (k = 36) used single-item assessments of sexual communication; the remaining studies measured communication with 2 to 5 items (k = 17), 6 to 10 items (k = 7), or more than 10 items (k = 8), and 3 studies did not specify the number of items they used. Among all studies, the primary design was cross-sectional (k = 64); 7 studies used a longitudinal design to examine parent-adolescent sexual communication as a predictor of later contraceptive or condom use.4,28,31,44
There was a small, significant overall weighted mean effect for the association between parent-adolescent sexual communication and safer sex behavior (r = 0.10; 95% CI, 0.08-0.13) (Figure 2).4,13- 19,21,23,28- 31,43,44,48- 83 Funnel plots of the effect sizes were symmetrical, and the trim-and-fill analysis suggested no adjustment to the mean effect size.84 This finding indicated no evidence of publication bias.
Although the overall association between communication and safer sex was positive and significant, there was considerable heterogeneity among the effect sizes (Q = 203.50, P < .001, I2 = 65.60). Thus, we examined the potential effect of several moderating variables. Studies were included in moderator analyses only if they had sufficient information to be analyzed. For example, when considering sex as a moderator, studies had to sample only boys, only girls, or both sexes but report separate analyses by sex (mixed-sex samples that did not separate analyses by sex could not be included because there was no way to distinguish the association between communication and contraceptive use for boys vs girls).
First, we examined moderators related to participant demographics and study design. As shown in the Table, there was significant moderation by sex (k = 48), with a stronger association between parent-adolescent communication and safer sexual behaviors with girls (r = 0.12) compared with boys (r = 0.04). The strength of this association was not found to differ significantly by adolescents’ age (k = 62) or ethnicity (k = 71) or by the location of the study (k = 70). In addition, effect sizes did not differ significantly when comparing longitudinal and cross-sectional study designs (k = 71).
Next, several aspects of sexual communication were examined as potential moderators (Table). Among studies that specified the source of communication (ie, mother vs father; k = 19), the association between sexual communication and safer sex behaviors among youth was significantly stronger for adolescents who had discussed sexual topics with their mothers (r = 0.14) vs those who had discussed sexual topics with their fathers (r = 0.03). In fact, communication with fathers was not significantly associated with safer sex behavior among adolescents across studies (r = 0.03; P = .46). The strength of the association between communication and safer sex behaviors did not differ significantly based on the topic of conversation (k = 71), the format of communication measurement (k = 70), or the number of items used to assess communication (k = 68).
Finally, 2 factors specific to the outcome of safer sex behavior were examined as moderators, including the type of safer sex behavior and timing of safer sex examined in each study. As shown in the Table, effects did not differ based on the type of safer sex behavior (k = 71), with similar significant associations found for contraceptive use (r = 0.09) and condom use (r = 0.12). Effects were also consistent across the measurement time frames (ie, lifetime, past 6 months, past 3 months, first sexual intercourse, and last sexual intercourse; k = 67).
Pooling data from 3 decades of research with more than 25 000 adolescents, this meta-analysis found a significant positive association between parent-adolescent sexual communication and safer sex behavior among youth. This effect was robust across use of condoms and contraceptives, cross-sectional and longitudinal studies, and younger and older samples. The strength of this association was moderated by sex of the adolescent and sex of the parent, with stronger effects for girls than for boys and for communication with mothers vs fathers.
First, the association between parent communication and adolescents’ contraceptive and condom use was significantly stronger for girls than for boys. This finding is consistent with past work showing that parents communicate more frequently with girls and are more likely to stress the negative consequences of sexual activity when discussing sexual activity with daughters compared with sons.21,23,30,48,85,86 If parents wish to exert a stronger influence on their sons’ safer sex practices, they may need additional training to change the frequency, content, and/or tone of the messages surrounding sexual activity that they communicate to boys.
Second, the association between communication and safer sex was also moderated by the sex of the parent. Specifically, adolescent communication with mothers was positively associated with use of protection, but there was not a significant association between father-adolescent communication and safer sex behavior. Across a variety of circumstances, men and boys are less verbally expressive, open to self-disclosure, and attuned to emotional and relational cues compared with girls and women.87 This difficulty in sharing emotional experiences or discussing potentially embarrassing relational topics may inhibit the ability of some boys and fathers to have open and intimate conversations about sexual health. It would be ideal to examine whether specific factors related to fathers’ communication—such as how often fathers are communicating with their sons or daughters and the specific content or comfort level of these conversations—might amplify the effect of communication on adolescents’ safer sex practices; unfortunately, there are currently too few studies of father communication to examine these fundamental questions. This remains a ripe area for future inquiry.
Taken together, our results confirm that, across more than 50 studies, parent-adolescent sexual communication is positively associated with adolescents’ use of contraceptives and condoms regardless of communication topic or format. However, this effect explained a relatively small proportion of the variance in safer sex behavior. Thus, results underscore the importance of understanding parent communication—likely a more distal predictor—in the context of more proximal factors that contribute to sexual decision making. Building on preliminary models of parent-adolescent communication,7,8,10 future theoretical and empirical work should examine how parent communication affects individual-level factors (eg, attitudes and self-efficacy)88 as well as couple-level factors (eg, partner communication and negotiation processes)37,89 and how these and other factors may mediate the association between parent-child communication and safer sex. In line with family systems theory,90 it is also possible that alternative parenting constructs, such as parent-adolescent relationship quality, parental monitoring, or the marital relationship itself, may interact with communication to predict sexual behavior among youth.39 Future work will benefit from in-depth analyses of the role that parental communication may play in adolescent sexual decision making within these multiple domains of influence.
Several methodological issues were identified in this literature review that may have obscured the detection of more robust effects and are worthy of future research attention. Addressing these study design issues may elucidate why several expected communication measurement characteristics (ie, topic and format)37,89 did not emerge as significant moderators of the association between communication and behavior.
First, many studies assessed sexual communication with unspecified parents. In these cases, it was not possible to know if youth were reporting communication with mothers, fathers, or both parents. Given the current findings that sex of the parent was a moderator, as well as previous research demonstrating more frequent communication about sexual issues with mothers than with fathers,91 the overall association we found may be driven mostly by communication with mothers. Additional research is needed to better understand this issue.
Second, more than half the studies used single-item assessments of parent-adolescent communication. This finding is not ideal from a measurement perspective because single-item assessments are unlikely to capture the nuance and complexity of the communication process. It is clear that the quality and timing of communication can have important implications for sexual decision making among youth.92 Given that many parents misjudge when their adolescents begin sexual activity, communication about sex may begin after the initiation of sexual activity and limit the potential effect of these discussions.22,93 To further our understanding of communication among adolescents and their parents, we need to use not only brief measures of the content or frequency of communication but also in-depth measures of the timing, tone, and style of these sexual discussions.10,36,94 This approach may require mixed-methods longitudinal studies in which quantitative reports are collected alongside qualitative interviews, perhaps using ecological momentary assessments to capture communication soon after it occurs. It would also be useful to obtain reports of communication from both adolescents and their parents to identify discrepancies in the frequency or quality of communication that each individual reports.13
Finally, of the 71 independent effects identified, only 7 used longitudinal designs. While there were no significant differences in the effect sizes drawn from cross-sectional vs longitudinal studies, additional work should use multi-wave longitudinal designs to determine the effect that timing of communication has on contraceptive behavior.
Results of this study confirm that parent-adolescent sexual communication is a protective factor for youth, and a focus on communication remains justified in future intervention efforts. Because conversations about sexuality can be uncomfortable or embarrassing for both parents and adolescents, educational efforts may be most successful if they provide clear, practical instruction and help parents optimize the timing and language used in their approach.25 In addition to formal intervention programs with parents, physicians and other health care professionals who interact with parents and youth are in a unique position to encourage healthy communication about sexual topics. Specifically, physicians can have clear and honest conversations about sexual health issues in professional settings to model sexual communication skills,95 perhaps helping families initiate these conversations. They can also urge parents and adolescents to have such conversations at home, as well as provide resources to parents on when and how to discuss sensitive sexual health topics.
This study fills a critical gap in the literature by meta-analyzing the association between parent-adolescent sexual communication and safer sex behavior among youth. Across more than 3 decades of research and 25 314 adolescents, this meta-analysis suggests that communication with parents—particularly among mothers and girls—has a protective effect on adolescent contraceptive and condom use. Further research using more sophisticated assessments, longitudinal designs, and mixed-methods approaches are needed to advance this literature and to better understand the effect parents have on the health of their adolescents.
Accepted for Publication: August 5, 2015.
Corresponding Author: Laura Widman, PhD, Department of Psychology, North Carolina State University, 640 Poe Hall, Campus Box 7650, Raleigh, NC 27695 (firstname.lastname@example.org).
Published Online: November 2, 2015. doi:10.1001/jamapediatrics.2015.2731.
Author Contributions: Dr Widman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Widman, Choukas-Bradley, Noar.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Widman, Choukas-Bradley, Garrett.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Widman.
Obtained funding: Widman.
Administrative, technical, or material support: Choukas-Bradley, Nesi, Garrett.
Study supervision: Widman, Noar.
Conflict of Interest Disclosures: None reported.
Funding/Support: This research was supported in part by grants R00HD075654 (Dr Widman), K24HD069204, P30AI50410, and UL1TR001111 from the National Institutes of Health. This work was also supported in part by the National Science Foundation Graduate Research Fellowship DGE-1144081 (Ms Nesi).
Role of the Funder/Sponsor: The funding sources provided support for the design and conduct of the study and the collection, management, and analysis of the data. They had no role in the preparation, review, or approval of the manuscript and decision to submit the manuscript for publication.
Disclaimer: The content, interpretations, and conclusions of this study are those of the authors and do not necessarily reflect the views of the National Institutes of Health or the National Science Foundation.