Abstract
Adolescents worldwide are at high risk for adverse consequences of sexual activity, including HIV, other sexually transmitted diseases, and unintended pregnancy. Effective intervention strategies are needed to address this risk. This article discusses the advantages of the reasoned action approach for developing such strategies, including the ability to integrate population-specific qualitative information with the approach to develop an intervention that is both theoretically grounded and culturally appropriate. It also describes an application of the approach in developing “Let Us Protect Our Future,” a culturally appropriate HIV risk-reduction intervention for adolescents in South Africa, where sexually transmitted HIV infections are having an especially devastating impact. The results of a randomized controlled trial revealed that grade 6 students in schools that received the intervention were less likely to report having sexual intercourse, unprotected sexual intercourse, and multiple sexual partners during the 12-month follow-up period than their counterparts in control schools.
Keywords
In the United States, adolescents are at high risk for sexually transmitted diseases (STDs) (Eng and Butler 1997; Weinstock, Berman, and Cates 2000), HIV infection, and unintended pregnancy (Santelli et al. 2004). But sexual risks are not limited to adolescents in the United States, for it is widely recognized that worldwide adolescents are at risk for adverse consequences of sexual activity (Joint United Nations Program on HIV/AIDS [UNAIDS] 2006). Strategies are needed to address these risks, and in this article, I describe the use of the reasoned action approach to develop and evaluate such strategies.
The Search for a Theory
My colleagues and I have conducted a program of research on HIV risk-related behavior that is based on the reasoned action approach. I began this research with an applied interest. I wanted to reduce the impact of the HIV epidemic, particularly on low-income African American populations. When I initiated my research in the late 1980s, few studies had developed and tested HIV risk-reduction interventions, and none had tested such interventions for African American populations. I began my research with the assumptions that a theory-based intervention would be more effective than an intervention that was not theory-based, and that an intervention that was culturally and developmentally appropriate would be more effective than one that did not have these qualities. The question was, Which theory should I use to develop such an intervention?
I searched the literature but could not identify any unique theory of the behavior of low-income African Americans relevant to sexual risk behavior and changes in such behavior. As a social psychologist, I then looked closely at the social psychology literature to identify a relevant theory. A number of theories seemed applicable. For instance, Festinger’s (1957) theory of cognitive dissonance might be employed to arouse dissonance by highlighting discrepancies between people’s advocacy for safer sex and their own past actions as a way of motivating them to engage in safer sexual behavior in the future (Aronson, Fried, and Stone 1991). Another candidate was McGuire’s (1985) model, which highlighted the importance of source factors, message factors, channel factors, receiver factors, and target factors as keys to the efficacy of persuasive communication. However, the theoretical framework that struck me as most useful to HIV risk reduction among African Americans was the theory of reasoned action (Fishbein and Ajzen 1975) and its extension, the theory of planned behavior (Ajzen 1991). These theories collectively have been labeled the reasoned action approach (Fishbein and Ajzen 2010).
In applying any theoretical approach to HIV risk reduction for African Americans, it was important to be mindful of cultural appropriateness (J. Jemmott and Jones 1993). The historical and contemporary experiences of white Americans compared with African Americans and members of other racial/ethnic groups have been markedly different, and accordingly, different underlying beliefs may motivate these groups’ HIV risk-related behaviors. There is the danger that a lack of understanding of these beliefs could lead to a misinterpreting of the behaviors of African Americans and members of other racial/ethnic groups not only in the United States (which was my initial concern) but also abroad. Of course, there is an extensive literature on the social psychological factors that influence behaviors, but for the most part, the participants in those studies were white Americans. Little empirical research had investigated the factors that underlie the behavior of people from other groups, including African Americans. Findings from empirical research on white populations may not generalize to African Americans and other populations. In a similar vein, the issue of cultural appropriateness arises in the extent to which the theories that have been developed based on research with primarily white Americans have application to the behavior of African Americans and members of other groups.
The Advantages of a Reasoned Action Approach
Certain features of the reasoned action approach made it an excellent choice as a theoretical framework for HIV risk-reduction research in a diversity of populations. First, the reasoned action approach has been used successfully to explain a broad array of health behaviors (Fishbein and Ajzen 2010). Although it might be argued that the approach would be inapplicable to sexual behavior inasmuch as such behavior is often seen as not “reasoned,” the approach has been applied successfully to sexual behavior (J. Jemmott et al. 2005; J. Jemmott, Jemmott, and Fong 1992, 1998, 2010; J. Jemmott et al. 1999), and several meta-analyses confirm the predictive power of the approach with respect to condom use (Albarracín et al. 2001; Armitage and Conner 2001; Sheeran and Orbell 1998; Webb and Sheeran 2006). Second, the reasoned action approach is flexible in that it allows for different attitudinal, normative, and self-efficacy/control determinants of intentions and behavior in different populations. It therefore can be adapted to explain different behaviors and can be applied to different people. The approach explicitly states that the predictors of behavior may vary depending on the behavior and the population. Finally, the approach assumes that the relevant predictors can be identified through qualitative research conducted on the population. This serves to tailor the theoretical measurements to the population (for more on tailoring, see Curtis, this volume). The approach may not include all of the variables that are important to a specific behavior in a specific population, but it provides a framework within which such variables can be understood and incorporated as precursor determinants of the three basic theoretical mediators.
In the reasoned action approach, behavioral intentions are the main determinants of behaviors, and intentions are determined by attitude, normative pressure, and perceived behavioral control or self-efficacy regarding the behavior. Thus, adolescents should intend to use condoms if they evaluate condom use positively, if they believe significant others think they should use condoms, and if they feel confident in their ability to use condoms. A valuable feature of the approach is that it directs attention to why people hold specific attitudes, subjective norms, and perceived behavioral control or self-efficacy. Behavioral beliefs about the consequences of using condoms determine attitudes toward using them. Normative beliefs about important referents’ approval or disapproval of condom use determine subjective norms. Control beliefs about factors that facilitate or inhibit condom use determine perceived behavioral control or self-efficacy.
It might be argued that several other factors not included in the reasoned action approach may affect behavior, for instance, age, poverty, sex, culture, gender-role expectations, parental monitoring and supervision, parent-child communication, religiosity, and alcohol and drug use. The reasoned action approach can accommodate these precursor or external variables. The effects on intention and behavior of variables external to the theoretical framework are mediated by their effects on the attitudinal component, the normative component, the perceived control component, or all three. In other words, external variables may affect the beliefs that are a part of the theory and, through a mediation chain, influence behavior. In this sense, the theory offers a clear prescription for intervention, which would be an external variable in the language of the theory. Reasoned action interventions target the salient beliefs and, through a mediation process, affect behavior.
The theory also offers a strategy for identifying relevant beliefs. Thus, one does not identify the beliefs by looking in the literature or by theorizing and hypothesizing about which beliefs might be important. Rather, one conducts qualitative research with the population to identify the population-relevant salient beliefs, which may differ for different behaviors and different populations. This qualitative research can involve the analysis of information from key informant interviews; elicitation surveys; focus groups; or even essays, journals, and poems, depending on the cultural context (for more on elicitations, see Middlestadt, this volume). By identifying and targeting salient beliefs, a reasoned action intervention may change attitude, subjective norms, and perceived self-efficacy, which would change intention, which, in turn, would change behavior.
Perhaps most important, the reasoned action theory allows the relative predictive power of the attitudinal, normative, and control components of the theory to vary from population to population. Thus, the relevant predictors of a behavioral intention might be different in middle-class, white college students (the population on which the largest number of studies have been conducted) compared with low-income African American women or compared with Xhosa-speaking adolescents in South Africa. In some populations, the attitudinal component may be most important; in others, the normative component may be most important. In still others, the self-efficacy component may be most important, but the theory should have high predictive value in each of these populations.
The Process of Intervention Development
In pursuing a reasoned action approach to intervention development, we employed a series of research phases. First, we conducted qualitative research to identify the target population’s salient behavioral beliefs, normative beliefs, and control beliefs relevant to sexual risk behavior and cultural and contextual factors associated with such behavior. Second, we used the information about the salient beliefs to develop a questionnaire to examine whether the beliefs are empirically related to intentions and behavior. Third, we integrated the results of the questionnaire and qualitative research, including the cultural and context factors, with the theoretical framework to design an intervention to affect the salient beliefs while taking cultural and contextual factors into account. Fourth, we pilot tested the intervention with the target population to examine the acceptability, feasibility, and logistics of implementing the intervention and the likelihood that it would affect the targeted beliefs in a larger trial. This may have involved more than one pilot test, as the intervention protocol was revised before it was fully developed.
Fifth, we tested the effects of the intervention using a randomized controlled trial, which provides the most scientifically valid evidence on the efficacy of an intervention. Commonly, there are three broad research questions of interest: (1) Was the intervention efficacious? (2) With whom was it most efficacious? and (3) Why was it efficacious or not efficacious? We followed all of these phases for each test of the intervention. In addition, if the intervention was efficacious, we tried to disseminate it so that it was available to those who could put it to use. There are also research questions that emerge with regard to dissemination of efficacious interventions (J. Jemmott, Jemmott, and Fong 2010). For instance, is the intervention effective when implemented under more real-world circumstances, not in a tightly controlled efficacy trial? How much training is required of educators to implement the intervention? What determines whether an efficacious intervention is widely adopted by potential end users?
Reasoned Action Interventions in South Africa
A project that we recently completed in South Africa is an excellent illustration of the use of the reasoned action approach in a cultural context that is quite different from that of the United States. South Africa has the largest number of persons living with HIV/AIDS in the world, an estimated 5.5 million (UNAIDS 2006). About 18.8 percent of South Africans ages 15 to 49 are living with HIV (South African Department of Health 2006), and new HIV infections are being driven by the high incidence in those ages 15 to 24. It is estimated that more than one-half of South Africans who were 15 years old in 2006 will not survive to age 60 (Dorrington et al. 2006). Young adolescents, before or just after the initiation of sexual activity, are important intervention participants, because they are highly vulnerable and have not established habitual patterns of sexual behavior.
Before conducting the South African project, we completed a series of studies in the United States. Over more than 15 years, we developed and demonstrated the efficacy of HIV risk-reduction interventions in reducing sexual risk behaviors in African American and Latino adolescents and African American women. We pioneered the use of “placebo attention” control groups, using an active non-HIV/STD intervention to guard against Hawthorne effects (J. Jemmott, Jemmott, and Fong 1992)—an approach that is standard practice in all of our intervention trials. Our interventions have targeted a variety of sexual risk outcomes, including not only condom use but also abstinence and delayed initiation of sex (J. Jemmott, Jemmott, and Fong 1998, 2010; Villarruel, Jemmott, and Jemmott 2006) and not only self-reported sexual behavior but also biologically confirmed STDs (J. Jemmott et al. 2005; L. Jemmott, Jemmott, and O’Leary 2007). Besides testing the efficacy of interventions, we have examined several practical issues relevant to dissemination, including the effects of the gender of the intervention facilitator (J. Jemmott et al. 1999; J. Jemmott, Jemmott, and Hacker 1992), the race of the facilitator, the sexual composition of intervention groups (J. Jemmott et al. 1999), and the age of facilitators (J. Jemmott, Jemmott, and Fong 1998).
In South Africa, we sought to develop a theory-based culturally appropriate intervention to reduce sexual risks among young adolescents (J. Jemmott et al. 2011, 2010). The study was a cluster-randomized controlled trial with schools randomized to an HIV risk-reduction intervention or a health promotion control intervention for grade 6 students (or learners, as they are called in South Africa) (J. Jemmott et al. 2011, 2010). The study included schools from an urban township, Mdantsane, and a nearby semirural settlement, Berlin. This was a case where we knew nothing about the culture yet wanted to develop an efficacious sexual risk-behavior intervention.
First, we conducted extensive preliminary research. This included focus groups with adolescents, parents, and teachers; meetings with school principals and heads of school governing boards; and interviews with key informants, including physicians, government officials in the ministries of health and education, councilmen, and representatives of nongovernmental organizations (NGOs). South African researchers, health officials, school officials, and representatives of NGOs were involved in the planning of the study from the beginning. We conducted twelve focus groups with Xhosa-speaking grade 6 students and parents and teachers of grade 6 from two schools in Mdantsane and one school in Berlin. We trained facilitators from a local NGO who had experience working with Xhosa adults and adolescents to conduct the focus groups. The focus groups helped us to identify the salient beliefs, cultural factors, and contexts that are relevant to HIV/STD sexual risk behavior among Xhosa-speaking adolescents. We held seven focus groups with sixty-nine adolescents (61 percent female and 39 percent male) 10 to 15 years of age (mean age = 12.6 years). We conducted four focus groups with thirty-four parents and one focus group with twelve teachers. The participants in the adolescent focus groups suggested that we name the project “Let Us Protect Our Future.” The focus groups also revealed several beliefs about sexual risk behaviors. Some were similar to beliefs seen in other populations, and some were unlike those we had previously encountered. Participants said that “condoms destroy the sensation of meat to meat or flesh to flesh” and that you “can’t eat sweets with [the] paper still on.” These are similar to what we call hedonistic beliefs (J. Jemmott et al. 1992), or beliefs about the negative consequences of condom use for sexual enjoyment. Other common beliefs were “condoms are too big for young boys,” “condoms prevent AIDS,” and “peers would approve of sex.” However, participants also said that “by having sex, you can pass on your seed,” “witchcraft can cause AIDS,” “not doing your duty by your ancestors can cause AIDS,” “sex with a virgin can cure AIDS,” and “men cannot say no to sex.” The focus groups also revealed the risk that adolescents (particularly girls) faced if they were home alone or were en route to and from school, which were situations in which they might be accosted by older men.
In addition, we conducted extensive pilot tests of the study questionnaire and held debriefing focus groups so participants could comment on the questions. Finally, we pilot tested the questionnaire in Xhosa with sixty-four grade 6 students (mean age = 12.4) from Mdantsane and Berlin. We wanted to determine whether adolescents understood the questions and response choices and to solicit their recommendations regarding question phrasing. Two versions of the questionnaire were pilot tested: a self-administration and a “read-aloud” administration. In the read-aloud administration, the data collector read each question and response category aloud. The participants had the questionnaire before them and completed the questions as the data collector read to them. We tried this procedure because of low literacy rates in English and Xhosa in the population. (English is the language of instruction from grade 4 onward, but Xhosa is the home language of the children.) We found that children had difficulty reading English, and although they understood Xhosa, they were not accustomed to or skillful at reading Xhosa. The read-aloud administration went well. It gave participants a better understanding of the questions and the response choices. The participants noted that this method was easier, and we found that the data were more complete and internally consistent. Therefore, we employed this method in the main trial.
We conducted two additional adolescent focus groups with twenty grade 6 students (mean age = 11.9) from Mdantsane and Berlin to address some specific issues related to curriculum development. We developed drafts of the “HIV/STD Risk-Reduction and Health Promotion Control” curricula based on the information gathered in the focus groups, meetings, and questionnaire sessions; input from South African coinvestigators and consultants; and our previous experience in developing curricula for adolescents. Each curriculum consisted of twelve 1-hour modules implemented over six sessions. Although the HIV/STD risk-reduction intervention retained core elements (negotiation and condom use self-efficacy and skill-building activities) of previously published interventions—namely “Making a Difference,” “Making Proud Choices,” and “Be Proud! Be Responsible!”—new activities, themes, supporting materials, and messages were developed to ensure that each intervention was contextually, developmentally, linguistically, and culturally appropriate for Xhosa adolescents.
We conducted three pilot tests of the interventions. The first was held in Mdantsane with thirty English-speaking grade 6 students (twenty-two girls and eight boys), 11 to 13 years of age (mean = 11.6), to determine whether adolescents understood the curricula and to assess their reactions to the activities. The participants were identified by their school’s principal as among the best grade 6 English speakers in the school. Team members who had extensive experience implementing HIV/STD and health-promotion curricula in the United States implemented the interventions in English. One pair of Xhosa-speaking male and female adults observed the implementation of each intervention and provided assistance with translation and interpretation. These adults later served as cofacilitators in Xhosa-language pilot tests of the curricula. The learners were extremely involved and actively participated in all activities. Attendance was excellent: only four absences among the thirty participants in the six-session interventions, and all returned to complete the postintervention questionnaire. Both interventions involved homework assignments to increase parent-child communication about the issues in the particular curriculum and to inform parents about what their children were learning. Inasmuch as sexual matters are taboo within Xhosa culture and we were concerned about parents’ reactions to some of the content of the curricula, the last homework assignment included questions to assess parents’ reactions to the interventions and the project. Twenty-seven of thirty learners returned their parents’ responses, and all responses were very positive.
We revised the curricula based on the pilot, translated them into Xhosa, back-translated from Xhosa to English, pilot tested them in Xhosa in Berlin and Mdantsane, and delivered them in Xhosa in the main trial. We trained the four Xhosa-speaking adults who had observed the interventions implemented in English to serve as cofacilitators in a pilot test of the interventions in Xhosa. We conducted this pilot with forty-three learners (twenty-one girls and twenty-two boys), 11 to 15 years of age (mean = 11.8 years), in Berlin whom we randomized to the HIV/STD risk-reduction or health-control intervention. Attendance was excellent: all participants attended all the intervention sessions and returned for the posttest. The responses of the participants and their parents once again were very positive. We implemented the final pilot test of the interventions in Xhosa in Mdantsane with forty-three learners (twenty-three girls and twenty boys), 11 to 15 years of age (mean = 12.7). We randomized the participants to three groups: an HIV/STD risk-reduction group and one of two health-promotion control groups. Attendance was excellent. One participant missed five sessions of the health-promotion intervention because of illness. There was only one other absence from any of the six sessions in the three groups. Both interventions went extremely well, and parent responses were very positive.
In addition to conducting formative research with the study population, we assembled a local Community Advisory Board (CAB) to advise us regarding the design and implementation of the study and the interventions. The CAB included parents, teachers, physicians, councilmen, representatives of the ministries of health and education, and representatives of NGOs that address the needs of adolescents. The CAB advised us on the sensitivity; cultural relevance; gender relevance; and contextual appropriateness, feasibility, and acceptability of the interventions, assessments, interpretation of the findings of preliminary studies, and dissemination of the findings. They provided suggestions and feedback on recruitment and retention strategies, intervention protocols, measures, and staff recruitment and hiring efforts. Their suggestions were incorporated into final versions of the intervention, protocols, and measures and helped to ensure that the study was culturally appropriate and acceptable to Xhosa adolescents and their parents. We held regular meetings of the CAB and met individually with members often to provide updates on the progress of the project.
We began enrolling participants in the main trial in October 2004. The objective of the trial was to test the efficacy of the HIV/STD risk-reduction intervention in reducing HIV/STD risk behavior compared with the health-promotion control intervention in a sample of Xhosa-speaking grade 6 students in Mdantsane and Berlin. We utilized a cluster-randomized controlled trial design. Schools were eligible to participate if they had grade 6 students and served the general population of learners, not just those with learning disabilities. We reasoned that specifically tailored interventions and specifically trained facilitators would be required to address adequately the needs of children with learning disabilities. One school that exclusively served children with learning disabilities was ineligible, leaving thirty-five eligible schools: twenty-six in Mdantsane (urban) and nine in Berlin (rural). All agreed to participate in the trial.
We created seventeen matched pairs of schools that were similar in number of grade 6 students, number of classrooms, and number of classrooms with electricity—the latter a proxy for poverty. In addition, we created separate matched pairs involving the rural schools and the urban schools. We randomly selected nine matched pairs: seven matched pairs involving urban schools and two matched pairs involving rural schools. Within each pair, we randomized one school to the HIV/STD risk-reduction intervention and the other to the health-promotion control intervention.
Grade 6 students at the eighteen selected schools who had signed parent/guardian consent forms were eligible to participate. To decrease self-selection bias, school administrators, potential participants, and recruiters were blind to the specific intervention to which the school had been randomized, and recruiters followed a common standardized scripted recruitment procedure at the intervention and control schools. To recruit the participants, project staff made announcements to the grade 6 students while at school. They distributed parent/guardian consent forms and cover letters that introduced the study and requested that the parents/guardians read and sign the consent forms and that the learners return them to the school. We also invited parents/guardians to meet with project staff at the schools to ask any questions or express any concerns regarding the participation of their children in the study. Learners also provided assent to participate in the study. We held the interventions and data collection sessions at the learners’ schools during the extracurricular period at the end of the school day. Data collectors who were blind to the participants’ intervention condition implemented the read-aloud procedure, and participants completed confidential questionnaires before; immediately after; and 3, 6, and 12 months after the intervention. Learners who completed the preintervention questionnaire and attended session 1 of the intervention were enrolled in the trial. As compensation for participating, we gave the learners notebooks, pens, and pencils for the postintervention questionnaire and the 3-month follow-up, T-shirts displaying the project logo for the 6-month follow-up, and backpacks displaying the project logo for the 12-month follow-up.
HIV/STD risk-reduction intervention
We designed the HIV/STD risk-reduction intervention to address the theoretical variables that influence abstinence and condom use. Specifically, the intervention goals were to (1) increase HIV/STD risk-reduction knowledge, (2) enhance behavioral beliefs supporting abstinence and condom use, and (3) increase skills and self-efficacy to negotiate abstinence and condom use. This included addressing the unique cultural beliefs. For instance, the intervention noted that the cultural value of passing on your seed does not have to mean having children before you are ready to support them. The intervention was highly structured and was implemented by male and female Xhosa-speaking cofacilitator pairs who used standardized intervention manuals.
Several features, including length, target age, and prevention message, make “Let Us Protect Our Future,” the HIV/STD risk-reduction intervention, quite distinct from our previous intervention curricula. “Let Us Protect Our Future” was designed to be appropriate for the context of adolescents in Mdantsane and Berlin and thus took into account Xhosa culture and the transformations of the culture that were occurring or had occurred in the urban township setting of Mdantsane at the time of this study. For example, in the area where we conducted our research, people do not generally refer to genital organs or sexual activity by their Xhosa names. Indeed, Xhosa words for sexual topics and private parts are considered obscene. We wanted to be faithful to the culture, but we also wanted to use language that adolescents would understand. We found that often the adolescents’ understanding of Xhosa was not very good, but neither was their command of English. We also found that people were more comfortable when we used English equivalents in relation to sexual organs. Therefore, we addressed language issues by “Xhosalizing” certain words or creating a hybrid of both English and Xhosa to maximize both understanding and comfort. In discussing STDs, we used English words because participants from our focus groups were more familiar with them. In discussing body parts, we mixed English and Xhosa. We also had to take into account the fact that in Xhosa culture women should not discuss certain issues with boys, and men should not discuss certain issues with girls. These are largely issues concerning puberty covered in the first intervention session. To ensure that participants felt comfortable to express themselves freely, we separated them by gender. Girls could have fruitful conversations about topics such as their menstrual cycle with their female facilitator; boys could discuss issues such as wet dreams with their male facilitator. However, even though these activities were implemented in single-sex groups, both groups of boys and girls covered exactly the same content.
Another adaptation in “Let Us Protect Our Future” was the use of comic workbooks. Electricity was not available in many of the schools in which we implemented the interventions, which meant that we could not use videos and video clips, which are commonly employed in interventions in the United States. We therefore developed comic workbooks—six issues, one for each session—with a series of characters and storylines that were used throughout the intervention to address HIV/AIDS, stigma, pregnancy, the impact of risky behaviors on goals and dreams, abstinence, condom use, and coercive sex. The storylines were based on what teachers, parents, and adolescents in our focus groups determined were real issues facing many adolescents in Mdantsane and Berlin. The comic workbooks helped to increase risk perception, provided a sense of ease when discussing sensitive topics, and facilitated discussion in a nonjudgmental manner. Reading aloud the dialogue of the characters in the comic workbooks during each session helped prepare participants for unscripted role-plays in the final session, where they practiced communication, negotiation, and refusal skills and techniques related to abstinence and condom use. Because the characters in the comic workbook were faced with these issues, the participants had opportunities to read about and discuss how each character conducted himself or herself in the given situation. For example, one of the characters was a teenage mother who was forced to reevaluate her goals and dreams for the future. Another character felt pressured by her boyfriend to have sex. These are real-life issues faced by Xhosa adolescents.
We also developed several other new activities for this setting. For example, in the “Long Walk Home Activity,” a maze handout was used to help participants first identify risky situations that they may encounter en route home from school and then trace the safest path home in an attempt to avoid dangers. In a similar vein, the comic workbook included a storyline to heighten participants’ awareness about inappropriate romantic relationships, sexual coercion, cross-generational sex, and sex for gifts. This was complemented by an activity on characteristics of good partners and bad partners.
Hats have special significance in Xhosa culture. The “Hat Activity” used images and color to invite participants to consider the many roles or “hats” one wears in life and to address self-pride and choosing one’s own role. Facilitator-debriefing reports indicated that most participants elected to wear their hats as they walked home following the session. We developed a “Doll Activity” to address attitudes toward women and instill that the clothing a woman wears is not a legitimate reason to attack her or to infer she has bad character. Participants in single-gender groups used dolls with changeable clothing to express their views on how young women dress in music videos, for example. Participants considered whether a girl who dresses in a sexy way is asking for sex or deserves to be forced to have sex.
Finally, the logo that we created for the project is a unique cultural adaptation. It is an image of a shield that protects the adolescents and their families. The logo features some of the comic characters, including the mother, father, sister, brother, older sister, and teacher, with the shield behind them. Participants from the focus groups were involved in suggesting an appropriate logo for the project. They were extremely excited to be involved in the creative process of the project, to the extent that they even suggested names for the characters, including the main characters: Khusela, meaning “Protect”; and Nangamso, meaning “Future.” These names dovetailed nicely with the project motto: “Let Us Protect Our Future.” The project motto was integrated throughout the intervention as a way to generate enthusiasm for the content.
We employed take-home assignments to enlist parents’ help in empowering their children to reduce their risk of sexually transmitted HIV infection and to ensure that parents were aware of the nature of the intervention. We sought to build enthusiasm and support for the project and comfort in and motivation for talking to their children about HIV, safer sex, and abstinence strategies; reduce barriers to talking to their children about these issues; and build their knowledge of HIV/AIDS transmission and risk-related behaviors. The focus groups with parents and with adolescents had revealed that they wanted to discuss sex, but “didn’t know how.” The assignments included a discussion of some of the concerns Xhosa parents have for their children and the cultural taboo regarding talking about sex, especially with their children. It also included a discussion on the barriers to and skills needed for talking to their children about sex (e.g., listening skills and teachable moments).
The health-promotion control intervention
The schools randomly assigned to the control condition received a 12-hour intervention focused on the health problems of the Xhosa population. This intervention was similar in structure and process to the HIV/STD risk-reduction intervention, except with different content. It had the same number of sessions, including an equal number of single-gender modules, led by Xhosa-speaking adult male and female cofacilitators. As in the HIV/STD risk-reduction intervention, there was a parental component of this intervention. The health-promotion intervention focused on behaviors associated with risk of heart disease, diabetes, high blood pressure, certain cancers, alcohol and drug abuse, and motor vehicle accidents—leading causes of morbidity and mortality among South Africans (Alberts et al. 2005; Asfaw 2006; Katz 2005; Steyn et al. 2006; Vorster et al. 2005). Participants were taught that these problems could be prevented by changing personal behaviors, including eating habits, physical activity, dental hygiene, cigarette smoking, and alcohol and drug use. The intervention also covered basic first aid and traffic safety. The health-promotion control intervention was designed to be as valuable and enjoyable as the HIV risk-reduction intervention (J. Jemmott et al. 2011).
Facilitators and facilitator training
Male-female cofacilitator pairs from the community implemented both interventions. We trained forty-three adults (twenty-one women and twenty-two men) 27 to 56 years of age, with a mean age of 42. Their median education was a bachelor’s degree; 65 percent had worked as teachers, and 63 percent had previously taught HIV education. We sought several qualities in facilitators: good oral communication skills, experience working with adolescents, the ability to follow written and oral directions, amenability to being evaluated so that the intervention process could be analyzed, and interpersonal flexibility. Facilitators needed to be dependable, personable, outgoing, and comfortable teaching sensitive health topics (e.g., breast self-examination and condom demonstrations). We hired individuals who had the basic skills to implement either of the two interventions. After stratifying them by gender, age, teaching experience, and HIV teaching experience, we randomly assigned them to be trained to implement one of the two interventions. In this way, we randomized facilitators’ characteristics across interventions; hence, any effects of the interventions cannot be attributed to the facilitators’ preexisting characteristics (J. Jemmott, Jemmott, and Fong 1998).
The facilitators received eight full days of training in their assigned intervention. The training included an overview of the project; data on sexual risks of adolescents; a description of the theoretical framework of the intervention; background information necessary to implement the intervention; and effective facilitation techniques, including presentation style, group and time management, and nonverbal and verbal communication. During the training, the master facilitators from the English and Xhosa pilots modeled the intervention activities, and the facilitators learned their assigned intervention; practiced implementing it; received feedback from each other, the master facilitators, and investigators; and created common responses to potential issues that could arise during implementation.
An important aspect of the facilitator training involved Xhosa facilitators’ verbalizing every term that they had ever heard, in both English and Xhosa, for body parts related to the reproductive system. This was done so that facilitators would be comfortable with the terms when interacting with participants. In addition, suggestions that facilitators made regarding appropriate terms were incorporated into the curricula. For instance, for a Xhosa equivalent to the word penis, the facilitator trainees at first insisted on the polite, but semantically ambivalent, word umphambili (the front) rather than the colloquial incanca, suggested in adolescent focus groups. However, we eventually agreed on the “Xhosalized” or hybridized form, ipenis (a/the penis).
Another key part of the training was to show facilitators that the success of the program was largely dependent on the adolescents’ willingness to communicate within the group. In Mdantsane and Berlin, the cultural environment presented us with a challenge in that children are expected to defer to their elders and speak only when elders speak to them. At school, learners are required to stand up when they address a teacher. Topics related to sexuality are often avoided. In our focus groups, parents and teachers, for example, expressed general discomfort talking to children about reproductive health and sexual subjects. Teachers, similar to parents or elders in general, have an unquestioned authority and power over young people. This makes it difficult for young people to express their feelings and concerns within the teaching environment without fear of retribution. To make up for this, facilitators were given special training that specifically emphasized facilitation and group process over the didactic approach that prevails in classrooms in Mdantsane and Berlin. It was important for facilitators to understand that communication in a group environment is facilitated by both trust and confidentiality. We have found that small groups of nine to sixteen participants and circle-seating promote confidence and break down the participants’ fear of speaking freely.
Another important aspect of the training was its emphasis on the importance of fidelity to the intervention. The intervention is highly structured, and facilitators received detailed standardized manuals to which they were trained to adhere in implementing each intervention session. The training stressed that all facilitators of the same intervention must implement it in the same way to reduce the effects of idiosyncratic facilitator factors. We told the facilitators that if different facilitators did different things in their small groups, an adequate test of whether the interventions were effective would not be possible. Overall, the training sessions went well, and the facilitators received them enthusiastically. During the eight consecutive days of training, none of the trainees was absent on any day. The facilitators received certificates for completing the training.
The enthusiasm of the facilitators for the training was matched by the adolescents’ enthusiasm for participating. Of the 1,898 grade 6 students enrolled at the schools, 1,396 (73.6 percent) returned signed consent forms, 1,118 were eligible to participate, and 1,057 of these (94.5 percent) participated. This included 558 girls and 499 boys ranging in age from 9 to 18 years, with a mean of 12.4 (SD = 1.2). As shown in Table 1, 7.6 percent resided in Berlin, and the others resided in Mdantsane. Only 38.8 percent lived in a household with their father. All eighteen schools remained in the trial through its completion. All participants attended intervention session 1; attendance at sessions 2 through 6 ranged from 97.0 to 98.6 percent. Follow-up return rates were excellent: 1,029 (97.4 percent) completed the 3-month follow-up, 1,030 (97.4 percent) completed the 6-month follow-up, 1,022 (96.7 percent) completed the 12-month follow-up, and 1,043 (98.7 percent) attended at least one follow-up. The percentage that attended at least one follow-up did not differ between the HIV/STD risk-reduction (98.8 percent) and control interventions (98.6 percent). Attending a follow-up session was unrelated to gender, father’s presence in the household, residing in Berlin, or sexual behavior. However, although follow-up attendance was excellent irrespective of age, participants ages 14 to 18 (96.0 percent) were less likely to return for follow-up than were those ages 12 to 13 (99.2 percent) or 9 to 11 (99.2 percent, p = 0.003).
Sociodemographic Characteristics of Participating Grade 6 Students, by Intervention Condition at Baseline, Mdantsane and Berlin, South Africa, 2004–2005
Effects of the HIV/STD Risk-Reduction Intervention
The primary outcome for the study was incidence of unprotected sexual intercourse in the previous three months. On this binary outcome, participants’ responses were coded 1 if they did not have vaginal intercourse without using a condom in the previous three months and 2 if they did have vaginal intercourse without using a condom in the previous three months. “Vaginal intercourse” was explicitly defined in the questionnaire since it is well known that different people have different understandings of the terminology “having sex.” The unit of inference in this trial was the individual. In the primary analyses, the efficacy of the HIV/STD intervention compared with the health-promotion control intervention over the 3-, 6-, and 12-month follow-ups was tested using generalized estimating equation (GEE) models, properly adjusting for longitudinal repeated measurements on learners clustered within schools (Fitzmaurice, Laird, and Ware 2004; Liang and Zeger 1986). Robust standard errors were estimated, and an exchangeable working correlation matrix was specified. The models included time-independent covariates, baseline measure of the criterion, intervention condition, and time (three categories representing 3-, 6-, and 12-month follow-up). We estimated the average intervention effects over the three follow-ups constructed from appropriate “estimate” statements from fitted GEE models and Cohen’s d’s, calculated by transforming the odds ratios using the Cox transformation (Sanchez-Meca, Marin-Martinez, and Chacon-Moscoso 2003). Analyses were performed using an intent-to-treat mode, with participants analyzed based on their intervention assignment, regardless of the number of intervention or data-collection sessions attended. Analyses were completed using SAS V9.
Table 2 presents descriptive statistics for sexual behaviors by intervention condition and assessment period. Table 3 presents estimated intervention effects during the follow-up period and corresponding significance tests. GEE analyses revealed that a smaller percentage of participants in the HIV/STD risk-reduction intervention schools reported having unprotected vaginal intercourse in the past three months averaged over follow-up (2.22 percent) compared with their counterparts in health-promotion control schools (4.24 percent), controlling for baseline unprotected vaginal intercourse (p = .0125; d=.41).
Observed Percentages Reporting Sexual Behaviors, by Intervention Condition and Assessment Period, Mdantsane and Berlin, South Africa, 2004–2006
GEE Empirical Significance Tests and Effect Size Estimates for the Overall Intervention Effect (3-, 6-, and 12-Month Follow-Up Assessments), Adjusted for Baseline Prevalence, Mdantsane and Berlin, South Africa, 2004–2006
ICC = intraclass correlation coefficient. ICCs were estimated by fitting a model using the generalized estimating equations approach and assuming an exchangeable working correlation structure with a logit link, unadjusted for baseline, controlling for clustering of learners within schools (n = 18).
Estimate = odds ratio (OR) (intervention versus health control).
We also examined several secondary behavioral outcomes, including ever experiencing vaginal intercourse and recent vaginal intercourse, heterosexual anal intercourse, and multiple partners in the previous three months. We had intended to examine self-reported consistent condom use as a secondary outcome, but too few participants reported sexual intercourse at baseline to allow for meaningful analyses of changes over time in condom use. Averaged over follow-up, participants in the HIV/STD risk-reduction intervention were significantly less likely to report having vaginal intercourse (4.75 vs. 7.20 percent, p = .0225; d=.29) and multiple sexual partners (1.83 vs. 3.19 percent, p = .0183; d=.42) in the previous three months than were participants in the health-promotion control intervention, controlling for baseline prevalence.
At baseline, 3.3 percent of participants reported ever having vaginal intercourse. Boys were more likely than were girls to report sexual experience (6.5 vs. 0.5 percent, p < .0001). Although the percentage who were sexually experienced increased to 23.2 percent by the 12-month follow-up, the effect of the intervention on sexual experience was nonsignificant. In addition, although the participants in the HIV/STD risk-reduction intervention were less likely to report having heterosexual anal intercourse in the previous 3 months (1.98 percent) than were the participants in the health-promotion control intervention (3.34 percent), this difference was not statistically significant (p = .073; d=.31). Few participants (0.38 percent) reported engaging in same-sex sexual behavior at baseline, and only 3.97 percent ever reported such behavior by the 12-month follow-up. The Intervention-Condition × Time interactions were nonsignificant, indicating that the effects of the intervention did not significantly differ among the three follow-up assessments.
We conducted analyses of school-level intervention effects to confirm the robustness of our findings from individual-level analyses. We used meta-analysis methods (Borenstein et al. 2009) that allowed us to take into account the number of learners in each pair of schools, giving more weight to larger pairs (similar to weighing larger studies more than smaller studies in a meta-analysis). This analytic approach also allowed us to examine whether the intervention effect size varied among the pairs of schools, just as a meta-analyst might test whether effect sizes were heterogeneous among trials. After fitting GEE models as previously described, we constructed mean estimated proportions for each outcome for each school (n = 18). We treated the estimated intervention difference for each pair as a study and constructed corresponding pooled estimates of intervention differences and 95 percent confidence intervals based on the nine pairs using the “metan” command in STATA, assigning weights to each pair proportional to the size of schools within pairs. Finally, we constructed an eight-degree-of-freedom test of whether intervention effects were heterogeneous across pairs. Table 4, which presents the results of the meta-analyses, reveals that results at the school level were nearly identical to those at the individual level, though slightly more significant statistically. Although the intervention effect on anal intercourse only approached statistical significance in the individual-level analysis, the school-level analysis revealed that the HIV/STD intervention produced a significant reduction in anal intercourse compared with the control intervention. The test of heterogeneity of treatment effects among school pairs was not significant on any outcomes.
Meta-Analysis Estimates of Pooled Intervention Effects and Corresponding 95 Percent Confidence Intervals of Predicted School Estimates from Fitted GEE Models
Estimate = odds ratio (OR) (intervention versus health control). Predicted probabilities obtained from fitted GEE models, adjusted for clustering by school (column 2, Table 3). School summaries output to STATA Version 10 for computations of pooled intervention effects using “metan” command.
We also examined the effects of the intervention on theoretical mediator variables. As shown in Table 5, we found significant effects across a range of theoretical mediator variables targeted by the intervention, including mediators of abstinence and mediators of condom use. For instance, the intervention increased the behavioral belief that practicing abstinence would increase the likelihood of achieving one’s career goals. The intervention decreased the normative belief that parents approve of the adolescent’s involvement in sexual activity and increased self-efficacy to avoid risky sexual situations. The intervention also increased the hedonistic behavioral belief that using condoms would not interfere with sexual enjoyment, self-efficacy to use condoms, and condom-use knowledge. The intervention also reduced the cultural myths regarding HIV. Thus, the intervention had durable effects on the theoretical variables that it was designed to affect and had no adverse effects on theoretical mediators of abstinence or condom use.
Longitudinal Estimates of Mediator Outcomes over the Follow-Up Period, by Intervention
NOTE: The estimates were obtained by fitting GEEs and transforming the empirical linear predictors to estimated means. The p-values are from one-degree-of-freedom chi-square tests of contrast estimates for intervention effects over the follow-up period.
Discussion
The results of the South African Adolescent Project demonstrate that the reasoned action approach is robust. We developed the intervention based on the reasoned action approach, integrated with qualitative findings on South African adolescents. The results suggest that such a theory-based, contextually appropriate HIV/STD risk-reduction intervention delivered in schools can be effective in shaping the sexual behavior of young adolescents before or at the beginning of their sexual lives. Averaged over the three follow-ups, the intervention reduced, by approximately 50 percent, the percentage of adolescents who reported unprotected vaginal intercourse compared with a health-promotion control intervention. Moreover, the intervention’s efficacy did not differ significantly among the three follow-up assessment periods.
This study also revealed that averaged over the three follow-ups, vaginal intercourse with multiple partners was reduced approximately 50 percent by the intervention compared with the control. This is important because having sex with multiple partners, particularly concurrent partners, is believed to play a critical role in the spread of HIV (Halperin and Epstein 2004; Rosenberg et al. 1999). Although the intervention produced a significant reduction in self-reported vaginal intercourse in the previous three months, it did not delay sexual debut. This contrasts with results of a study with adolescents in Namibia (Stanton et al. 1998), in which the intervention delayed sexual debut but did not reduce recent vaginal intercourse.
The rates of sexual activity in this sample of young adolescents were relatively low, which presents challenges in achieving adequate statistical power. Despite this, we obtained significant effects on important sexual behavior outcomes. That school-level analyses confirmed the primary individual-level analyses strengthens confidence in the findings. In developing the intervention, we drew on qualitative research with the target population, and analyses would suggest that we were successful in operationalizing the constructs into intervention activities. Paralleling the significant effects on behavioral outcomes, there were significant effects on theoretical mediators of sexual risk behaviors, including behavioral beliefs, normative beliefs, and self-efficacy as well as measures of HIV risk-reduction knowledge and cultural myths targeted by the intervention. It is noteworthy that as a consequence of participating in the intervention, the adolescents decreased their belief that their parents would approve of their having sex, an effect that may have been facilitated by the take-home assignments the children did with their parents.
Some might reason that, because we applied a Western theoretical framework to behavior change in a sub-Saharan African country, intervention effects would be weak. However, the findings contradict this reasoning. The effects of the interventions were at least as strong as those in meta-analyses of interventions for adolescents in the United States (Darbes et al. 2008; Johnson et al. 2003, 2009; Mullen et al. 2002). The reasoned approach of integrating theory with qualitative information from the population, then, may yield an efficacious contextually appropriate intervention.
A limitation of the study is the reliance on self-reports of behavior. Using incidence of STDs or HIV as an outcome was not feasible in our population with a 96.7 percent rate of sexual inexperience. Another limitation is that the results may not generalize to all South African adolescents. Few studies have been conducted with South Africans as young as our participants. Three studies that included adolescents comparable in age to our participants revealed rates of self-reported sexual inexperience greater than (Flisher et al. 2006; Richter 2006) or similar to (Mathews et al. 2008) those we observed.
In conclusion, sexual transmission of HIV is a major risk faced by adolescents in sub-Saharan Africa, and interventions are needed urgently to reduce their risk. This study provides evidence of the utility of applying the reasoned action approach to develop an intervention to reduce sexual risk behaviors, particularly unprotected vaginal intercourse, vaginal intercourse, and having multiple partners, among young South African adolescents in the earliest stages of their sexual lives. Future research with more sexually experienced adolescents will have to explore whether such interventions can have an impact on condom use and STDs, including HIV. Nevertheless, the fact that previous studies employing the reasoned action approach have shown that efficacious interventions can be delivered by community-based organizations and can influence not only self-reported outcomes but also biological outcomes suggests the likely success of efforts to implement the intervention widely in the sub-Saharan African context, where HIV is having its most devastating impact.
Footnotes
Preparation of this article was supported by NIH grant R01 MH065867.