Correlates of Unprotected Vaginal Sex Among African American Female Adolescents: Title and subTitle BreakImportance of Relationship Dynamics FREE

SEXUALLY transmitted diseases (STDs) are a common source of adolescent morbidity, with their sequelae being especially problematic and costly in females.¹ Understanding why adolescents at risk for STDs engage in unprotected sex is an important aspect of developing and implementing prevention programs.^{1 - 2} Yet, this issue has not been sufficiently addressed for female adolescents, particularly those at greatest risk of STDs.

Because of a combination of biological factors (eg, cervical ectopy) and social factors (eg, greater prevalence of STDs among their sex partners), African American adolescent females are particularly likely to be infected with an STD.^{1 ,3 - 4} Despite reporting greater frequency of condom use than their white and Hispanic counterparts,⁵ African American female adolescents experience disproportionately high rates of STDs, including human immunodeficiency virus infection.^{1 ,6 - 8} High prevalence of STDs among African American adolescent boys, combined with a tendency of African American females to select African American male sex partners, particularly those who have had 4 or more partners in the past year, increases the likelihood of female African Americans encountering an infected male partner, which, consequently, increases their risk of STDs.^{9 - 10} Risk of STDs for female African Americans is further magnified by low income⁸ and living in urban areas, particularly urban areas of the southern United States.^{7 ,11 - 12}

To date, quantitative research identifying correlates of risk behavior in samples composed, at least in part, of female African Americans has focused on adults rather than adolescents. Although these studies^{13 - 24} span a broad range of populations, variables representing relationship dynamics (eg, length of relationship, time spent with partner, and who controls sexual decisions) typically have not been included as part of the investigation. Similarly, several studies^{25 - 27} have focused on the sexual risk behaviors of African American adolescents; however, these studies have not thoroughly investigated associations between STD risk behaviors and relationship dynamics. Yet, qualitative research^{15 ,28} has indicated that relationship dynamics have an important effect on adolescents' sexual risk behaviors.

Understanding relationship dynamics may be important for several reasons. For example, qualitative research^{14 - 15 ,28 - 29} suggests that gender roles affect which partner will control sexual decision making. Gender roles favoring male dominance in sexual decision making may pose barriers for females relative to negotiating sex or condom use with a male partner.^{3 ,13 - 15} This may be especially true with female adolescents, a population likely to be less experienced than women in the complex process of negotiating with sex partners.

Length of relationship and amount of time spent with the partner may be important indicators of familiarity and trust in the relationship. Increased familiarity and trust may lead to an illusion of partner safety, negating motivations to use condoms for disease prevention.³⁰ Likewise, partner resistance to condom use may be a primary barrier to avoiding unprotected vaginal sex (UVS) among females.^{13 - 19 ,24} Lack of communication between sex partners about issues such as preventing STDs and human immunodeficiency virus and having older partners may contribute to females' reports of male partners' resistance to condom use.^{31 - 33}

Past studies have also focused on identifying correlates of condom use rather than correlates of unprotected sex; these 2 outcomes are distinct. Although past studies^{5 ,20 - 21 ,34} have used various measures of condom use, when disease prevention is of interest, measuring unprotected sex may be superior to measuring how often condoms were used.³⁵ Although condom use is protective against STD infection, female adolescents might also protect themselves by avoiding sex. Studies examining only condom use do not capture this behavior. Whereas condom use is a measure of relative risk, UVS is a measure of absolute risk. For example, an adolescent who used condoms 90% of the time for 100 episodes of intercourse would have 10 risk occasions; however, an adolescent who used condoms 50% of the time for 10 episodes of intercourse would have only 5 risk occasions. Thus, assessing percentage of condom use would be a misleading measure of risk because it fails to capture the absolute number of possible exposures to an STD. Therefore, the criterion measure for the present study was unprotected sex rather than the percentage of times a condom was used, thereby avoiding multiple statistical biases and subsequent misinterpretations.^{34 ,36}

Quantitative research is needed to identify specific constructs capturing relationship dynamics associated with sexual risk behaviors, specifically UVS. Such research could lay the groundwork for future research designed to investigate culturally specific aspects of relationship dynamics. Because African American female adolescents are disproportionately affected by STDs, we chose this population for initial investigation of relationship dynamics.

The primary purpose of this study was to determine associations between measures of relationship dynamics and frequency of engaging in UVS among African American female adolescents at high risk of STDs. The study also determined associations between adolescents' perceptions about safer sex and frequency of engaging in UVS.

Between December 1, 1996, and April 30, 1999, project recruiters screened 1130 teenage girls in adolescent medicine clinics, health department clinics, and school health classes to assess eligibility for participating in a human immunodeficiency virus and STD prevention trial. The present study consists of those adolescents who completed baseline assessments. Adolescents were eligible to participate in the trial if they were African American females 14 to 18 years at the time of enrollment, were sexually active in the previous 6 months, and provided written informed consent. More than half of the teens were eligible (n = 609 [53.9%]). Of those adolescents not eligible to participate (n = 521), most (98%) were not sexually active. Of the 609 eligible adolescents, 522 (85.7%) agreed to participate in the study. Most eligible teens who did not participate in the study were unavailable because of conflicts with their employment schedules. The recruitment sites were in neighborhoods characterized by high rates of unemployment, substance abuse, violence, and STDs. The study protocol was approved by the Institutional Review Board Committee on Human Research before implementation of the study.

Data collection was conducted at the Family Medicine Clinic and consisted of 3 components: a self-administered survey, a structured personal interview, and collection of vaginal swab specimens. The self-administered survey was conducted in a group setting with monitors providing assistance to adolescents with limited literacy and helping to ensure confidentiality of responses.

Subsequently, adolescents completed a face-to-face interview that assessed sexual risk behaviors. The interview was administered by trained African American female interviewers in private examination rooms. This data collection strategy has been suggested as an appropriate method for obtaining reliable and valid data related to sexual behaviors in African American adolescent populations.³⁷ On completing the interview, adolescents were asked to provide a vaginal specimen for STD testing. Adolescents were reimbursed $20 for their participation.

To determine the criterion measure for the analyses, we calculated an index of UVS for a retrospective period of 6 months. This index was created by subtracting the number of times adolescents reported using a condom during vaginal sex from the total number of times they reported engaging in vaginal sex. The index was calculated for UVS with all steady partners in the past 6 months for those reporting sex with steady partners and for UVS with all casual partners (a nonsteady partner) in the past 6 months for those reporting sex with casual partners. Adolescents self-defined whether their partner was steady or nonsteady. Because previous research³⁸ has established consistent differences in protective behaviors for persons in steady vs casual relationships, we used separate models to predict UVS among those in steady vs casual relationships.

Potential correlates of UVS were selected using social cognitive theory. Previous research^{27 ,39 - 41} based on social cognitive theory indicates that adolescents' perceptions of their social and peer environments as well as their own level of risk are associated with their sexual risk behaviors. For example, perceived use of condoms by peers and perceived risk of STD infection have been associated with condom use and may also be associated with UVS. Single-item measures assessed peer environment and self-perceptions: (1) perception of fewer girlfriends using condoms, (2) greater belief that other methods of contraception prevent STDs, (3) greater belief in invulnerability to STDs, and (4) less belief that condoms protect from STDs. In addition, a scale was used to assess adolescents' perceptions about condoms detracting from sexual pleasure and romance (α = .80). This scale yielded a variable we labeled "personal barriers to condom use." Also, a scale was used to assess adolescents' pregnancy worry (α = .71). Based on previously demonstrated utility in related studies, we assessed adolescents' history of STD infection.

Social cognitive theory also posits that relationship dynamics between sex partners will have a large influence on their sexual risk behaviors.^{42 - 43} A variety of relationship dynamics were measured. Because some of these constructs are broad, scales were used to optimize measurement. These scales represent constructs previously associated with risky sexual behavior in related studies: (1) belief that male partner control is normative,^{14 ,17 ,20 ,24} (2) perceived partner-related barriers to negotiating condom use,^{13 - 20 ,24} (3) open communication with partner,^{32 - 33} (4) fear of condom negotiation,^{13 - 20} and (5) condom negotiation self-efficacy.^{23 ,27} Table 1 displays the scale measures and a sample item from each scale, the number of items composing each scale, their associated reliability coefficients, and descriptive statistics. High scores represent greater levels of the scaled construct. A more complete description of the scales can be obtained by contacting the corresponding author.

Remaining measures of relationship dynamics were amenable to assessment with single items, as follows: (1) "During the past 6 months, I have felt that I have control over condom use"; (2) "During the past 6 months, I have felt that my boyfriend controls whether or not we have sex"; (3) age difference between typical sex partner and the participant; (4) greater length of steady relationship; and (5) higher average number of hours spent each week with steady boyfriend. The latter 2 variables were not entered into the model for casual partners. In addition, 2 demographic variables were entered as correlates: age and enrollment in school.

Pearson product moment correlations were calculated to assess the strength and direction of the bivariate relationships. Hierarchical regression was used to assess the level of independent contribution for each of the significant bivariate correlates to explaining the frequency of UVS for steady and casual partners. Correlates were entered, by block, into the regression models using a forward stepwise procedure within blocks. The α criteria for entry and exit were set at .05 and .10, respectively. For the model assessing UVS with steady partners, 3 blocks were entered: age and school enrollment, self-perceptions, and relationship dynamics. Only the latter 2 blocks were used for the model pertaining to casual partners. The F statistic was computed to test the overall significance of the final models. Acceptance of statistical significance was based on α = .05. We chose to use a stepwise entry method because of the dearth of published studies investigating relationship dynamics and female adolescents' sexual risk behaviors.

The following hypotheses were tested:

Variables indicating lower perceived and behavioral control in relationships will be associated with more frequent UVS. Specifically, we expected more frequent UVS to be associated with (1) stronger beliefs favoring male control, (2) greater partner barriers to condom use, (3) less frequent sexual communication with partners, (4) greater fear of condom negotiation, (5) lower condom negotiation self-efficacy, (6) stronger belief that partner controls decisions about having sex, (7) stronger belief that partner controls decisions about condom use, (8) having older sex partners, and (9) longer length of relationship and greater average time spent with partner (tested only in adolescents in steady relationships).

A constellation of peer environment and self-perceptions will be associated with increased frequency of UVS. Specifically, we expected more frequent UVS to be associated with (1) less belief in condom efficacy for STD protection, (2) less worry about pregnancy, (3) perception of fewer girlfriends using condoms, (4) greater perceived invulnerability to STDs, (5) greater belief in the STD protective value of other contraceptive methods, (6) perceptions that condoms detract from sexual and romantic pleasure, and (7) an absence of STD history.

All participants (N = 522) were African American females aged 14 to 18 years. The mean (SD) age of the sample was 16.0 (1.2) years. Most participants (81.2%) were full-time students; 9.4% were part-time students, and the remainder were not enrolled in school. Less than one fifth of the sample (17.8%) reported having a paying job. One fifth of the sample reported ever having an STD, and 28.2% had positive test results (by DNA amplification assay or culture) for at least 1 of the 3 STDs assessed at baseline interview: gonorrhea, chlamydia infection, or trichomoniasis.

We used a recall period of 6 months to allow for ample time to represent the lives and relationships of the adolescents. Most participants (81.8%, n = 426) reported having sex only with a steady partner; 8.2% (n = 44) reported having sex with only a casual partner(s); and 10.0% (n = 51) reported having sex with steady and casual partners. Unprotected vaginal sex was assessed for adolescents reporting sex with steady partners (n = 426 plus n = 51) and with casual partners (n = 55 plus n = 51). The mean (SD) number of reported acts of UVS with a steady partner in the past 6 months was 4.5 (13.3). Of adolescents reporting vaginal sex with a steady partner (n = 477), 51% reported at least 1 occasion of UVS (range, 1-180 occasions). The mean (SD) number of reported acts of UVS with a casual partner in the past 6 months was 0.74 (2.5). Of adolescents reporting vaginal sex with a casual partner (n = 95), 25% reported at least 1 occasion of UVS (range, 1-20 occasions).

Significant bivariate associations between hypothesized correlates and UVS for adolescents in steady relationships vs casual relationships are displayed in Table 2. With some exceptions, the bivariate associations for adolescents in steady relationships were significant and in the predicted direction. Contrary to our expectation, worry about pregnancy was positively associated with UVS. Alternatively, few correlates were related to UVS for adolescents in casual relationships. Of importance, 2 variables that were not related to UVS with steady partners were strongly related to UVS with casual partners: adolescent girls' perceptions that their boyfriends control sex and that their boyfriends control condom use. Significant relationships for adolescents in casual relationships were in the expected direction, with one exception: adolescents reporting that they did not need to use condoms because they were using other methods of contraception were less likely to report UVS.

Table 3 displays results from the regression model predicting UVS with a steady partner. The final model comprised 7 variables, explaining 23.5% of the total variance associated with UVS (F_9,396 = 13.5; P<.001). Size of the β weights indicates the relative strength of each variable in explaining UVS frequency. Thus, average time spent with boyfriend was the most important correlate of UVS frequency, with adolescents spending more time per week with their boyfriend reporting more frequent UVS. Likewise, adolescents reported more frequent UVS if they had longer relationships and held normative beliefs favoring male control. These relationship dynamics significantly added to the amount of variance explained by the self-perceptions.

Table 4 displays results from the regression model predicting UVS with casual sex partners. The final model comprised 3 variables, explaining 21.9% of the total variance associated with UVS (F_3,86 = 8.0; P<.001). Adolescents perceiving greater levels of personal barriers to condom use reported more frequent UVS. Likewise, adolescents reported greater frequency of UVS if they had no history of STDs. One relationship dynamic remained significant in the multivariate model: perceiving that the male partner decides whether sex will occur.

Adolescents' perceptions, particularly their perceptions of relationship dynamics, played an integral role in explaining female adolescents' frequency of UVS with steady and, to a lesser extent, casual partners. Two correlates were particularly important for adolescents in steady relationships: length of the relationship and average time spent with the boyfriend each week. These findings suggest that African American adolescent females become less concerned about risk of STDs as familiarity with their male partner increases. Thus, our results extend previous research,³⁰ showing that familiarity fosters beliefs of STD safety for sex partners, to a sample of African American female adolescents. Not using a condom may be viewed as a demonstration of trust and commitment in a relationship. Alternatively, adolescents spending more time with their boyfriends may report more frequent UVS because they engage in sex more often.

Also, among adolescents with steady partners, those who more strongly believed that male control in relationships is normative reported more frequent UVS. Thus, adolescents who have normative beliefs favoring male control in a relationship may be at increased risk of STD infection.

Among adolescents who had sex with casual partners, a single measure of relationship dynamics was important: perceiving that "my partner decides if we have sex." Thus, male partners of these female adolescents may control whether sex occurs. Despite the potential transient nature of casual relationships, this finding indicates the relevance of relationship dynamics in determining the outcome of sexual negotiations.

The results of this study suggest an important role of relationship dynamics in female adolescents' sexual partnerships. For those in steady partnerships, dynamics relevant to emotions such as trust and commitment seemed to be associated with increased frequency of UVS. Alternatively, for those in casual relationships, a single dynamic relevant to power differences seemed to be an important correlate of UVS.

In addition to relationship dynamics, correlates related to the peer environment and self-perceptions were important. For adolescents in steady relationships, the contribution of the variable measuring perceived social norms toward condom use (perceived number of girlfriends using condoms) to the model was particularly prominent, suggesting the importance of perceived social norms in determining sexual risk behaviors for African American female adolescents. Finding that greater perceived invulnerability to STDs predicted more frequent UVS with steady partners was expected; however, the finding relative to pregnancy worry was contrary to our hypothesis. Nonetheless, given that adolescents engaged in UVS with their steady partners, the unexpected positive relationship of pregnancy worry with UVS was understandable if one considers that increased UVS may lead to increased pregnancy worry.

Among adolescents who had sex with casual partners, the perception that condoms detract from sexual pleasure and romance was associated with more frequent UVS. These adolescents may perceive the benefits of condom use as being minimal compared with the benefits of sex without using a condom.

Only one correlate was common to both adolescents in steady and adolescents in casual relationships. As hypothesized, ever having an STD decreased the likelihood of reporting UVS, regardless of type of relationship (steady or casual). This finding is consistent with previous research by Anderson and colleagues,²¹ who reported results from a nationally representative sample of females aged 17 to 44 years, that showed a significant and positive correlation between recent condom use and ever having an STD. Similarly, Roye⁴⁴ reported that having an STD was the best predictor of increased condom use among a sample of female adolescents. Our findings suggest that the experience of having an STD and the experience of being treated for an STD (including preventive education and counseling routinely provided by the clinical staff) may motivate adolescents to reduce their frequency of UVS. This finding should not be confused with apparently contradictory findings⁴⁵ showing high reinfection rates of STDs for female adolescents. Reinfection with an STD may be due less to frequency of UVS than to sexual mixing patterns^{9 - 10} or engaging in concurrent relationships.⁴⁶

These findings are limited by the validity of the self-reported measures, particularly possible underreporting of sex with casual partners. Furthermore, inherent limitations of the cross-sectional design preclude determination of causality. Thus, longitudinal research is needed to establish over time the stability of observed relationships. Also, we chose not to include measures of unprotected anal and oral sex in the analyses because of the low prevalence of the former⁴⁷ and the relatively low STD risk of the latter.⁴⁸ Thus, the findings can be generalized only to female adolescents engaging in UVS.

These findings contribute to an empirical understanding of UVS and can be useful in designing gender-relevant interventions aimed at reducing the incidence of STDs among high-risk African American female adolescents.⁴⁹ Clinician-delivered STD prevention messages to adolescents receiving general health examinations can enhance adolescents' condom use⁵⁰ and, consequently, may reduce risk of STD infection. In addition, other clinic-based STD prevention programs have demonstrated that brief counseling can reduce STD risk behaviors.^{51 - 53}

Alternatively, nonclinic-based interventions that incorporate multiple sessions in a small-group format have also demonstrated efficacy with predominately adolescent and adult females.^{34 - 35 ,54 - 55} Yet, efficacy trials have not been designed specially for African American female adolescents. Thus, our findings suggest implications for the design of small-group interventions that seek to reduce female adolescents' STD risk. Clearly, these programs could promote a multitude of skills. Based on findings of this study, clinic-based STD counseling and small-group interventions designed for high-risk African American female adolescents should emphasize several key points. In clinic-based settings, implementation of these strategies is limited by the time made available for clinician-client interactions.

Findings suggest that female adolescents being treated for an STD or who are at risk for STDs need to hear messages debunking the myth that familiarity qualifies a person to judge the STD safety of a male sex partner. This myth can be dispelled by briefly describing the asymptomatic nature of highly prevalent STDs (eg, genital herpes, most strains of human papillomavirus, and chlamydia) and by emphasizing that the male partner may not be aware that he has an STD. Clinicians can emphasize this point by assessing STD risk of their adolescent female clients based on American Medical Association Guidelines for Adolescent Preventive Services,⁵⁶ screening at-risk clients for STDs, and urging STD testing for the male partner of adolescent female clients at risk for STDs. The findings also suggest that portraying condom use as a normative behavior is likely to reduce adolescents' frequency of UVS. Thus, African American adolescent female clients need to know that condom use is common among their peers, with this population being more likely to report condom use at last intercourse than their white or Hispanic counterparts.⁵

African American adolescent female clients diagnosed as having an STD can benefit from this experience. Thus, clinicians and educators should view diagnosis of an STD as a window of opportunity to engage adolescents in a discussion of STD preventive strategies and attempt to modify adolescents' perceived peer norms regarding frequency of condom use. This point is especially important for clients with multiple infections, a clear signal for intensifying prevention efforts. Clients at risk for STDs may benefit from teaching efforts designed to reduce perceived invulnerability to STDs. These messages are most likely to be effective when delivered interactively.⁵¹

The important relationship dynamic for female adolescents was partner control over the decision to have sex. Thus, teaching female clients at risk for STDs how to negotiate whether sex will occur may prove valuable in lowering their frequency of UVS. Addressing personal barriers to condom use may also be an important means of reducing UVS with casual partners among African American female adolescents. These adolescents may benefit from counseling designed to enhance perceptions of condom use as part of a satisfying sexual experience. Also, as with steady partners, using the diagnosis of an STD as a teachable moment is likely to be a beneficial strategy to reduce UVS.

Accepted for publication March 28, 2000.

This study was supported by grant 1R01 MH54412 from the Center for Mental Health Research on AIDS, National Institute of Mental Health, Rockville, Md (Dr DiClemente); an Association of Teachers of Preventive Medicine/CDC STD Prevention Fellowship (Drs Crosby and Sionéan); and a postdoctoral fellowship from the Nell Hodgson Woodruff School of Nursing, Emory University, Atlanta, Ga (Dr Cobb).

We thank Susan L. Davies, PhD, University of Alabama, Birmingham (UAB) School of Public Health; Edward Hook III, MD, Division of Infectious Diseases, UAB School of Medicine; and M. Kim Oh, MD, Department of Pediatrics, UAB School of Medicine, for their assistance and input. We also thank Jane R. Schwebke, MD, Division of Infectious Diseases, UAB School of Medicine, for provision of cultures for Trichomonas vaginalis, and Kim Smith, MT(ASCP), for assistance and oversight of testing for Neisseria gonorrhoeae and Chlamydia trachomatis.

Reprints: Richard A. Crosby, PhD, Department of Behavioral Sciences and Health Education, Rollins School of Public Health of Emory University, 1518 Clifton Rd NE, Room 520, Atlanta, GA 30333 (e-mail: rcrosby@sph.emory.edu).