Introduction

More than three million children worldwide are infected with human immunodeficiency virus (HIV),>90% of whom live in sub-Saharan Africa [1]. As the epidemic matures and more children are accessing antiretroviral therapy (ART), children who were born with HIV are aging into adolescence in large numbers [2]. It is estimated that approximately 40% of the more>3 million HIV-infected children worldwide have passed their 10^th birthdays [3]. Normative developmental transitions make adolescence a challenging time [4] and adolescents with HIV have worse adherence to treatment than both adults and younger children [5], [6], [7], [8], [9]. Adolescents' adherence challenges endanger not only the wellbeing and longevity of the individual youth, but also threaten to spark a new generation of youth with drug-resistant HIV as HIV+ adolescents with poor treatment adherence engage in sexual activity and potentially transmit resistant virus [10], [11].

When today's adolescents were born, Botswana's HIV epidemic was notable for both the high prevalence of infection and the government's unprecedented proactive response [12], [13]. In 2001, public prenatal clinics in Botswana began to offer free HIV screening and interventions to prevent mother-to-child transmission of HIV (PMTCT). In 2004, the program introduced routine “opt out” testing for HIV, dramatically increasing successful PMTCT efforts among the approximately 37% of pregnant women who were HIV-infected [14]. The government-sponsored treatment program has allowed perinatally HIV-infected children to have successful long-term outcomes [15]. Today, thanks in large part to the successful PMTCT and treatment programs, adolescents comprise the majority of infected pediatric cohorts in Botswana. Countries in the region with less well-developed prevention and treatment programs, however, are similarly seeing increased numbers of perinatally-infected adolescents as their epidemics mature since a substantial minority of perinatally-infected children will survive even without treatment [2].

The threat of poor adolescent adherence to treatment success and the magnitude of the epidemic among perinatally HIV-infected adolescents have created a substantial public health challenge [16]. Responding to this challenge requires a deeper understanding of the developmental processes and potentially modifiable risk factors for treatment non-adherence among these adolescents. Social support [17], self-efficacy [18] and substance abuse [19], [20] have been shown to be important to adolescent adherence in the U.S., but lack locally-validated measurement tools (See Figure 1). Based on our clinical experience and knowledge of adolescent development, we hypothesized that five separate factors play important roles in making treatment adherence challenging for adolescents. These included family support/supervision, psychological reactance, perceived stigma, outcome expectancy, and future orientation (See Figure 1). Social support can be defined in many ways. Here, we define social support as material and emotional support received from the people whose support is considered most important to the adolescent, as this type of support has been shown to be relevant to adolescent treatment adherence [21], [22]. We define family supervision in a more narrow sense, specifically focusing on the extent to which medication-taking is facilitated and observed by family. Psychological reactance is a strengthening of resistance to a desirable action in response to perceived threats to behavioral freedom. It peaks during adolescence and leads to increases in risk-taking behaviors [23].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Psychosocial Factors to Be Measured.

https://doi.org/10.1371/journal.pone.0109302.g001

Tools to measure these factors have similarly not been culturally-adapted and validated for use among adolescents in Botswana. We aimed to culturally-adapt and establish the construct validity of tools designed to measure these constructs. Construct validity is the extent to which a test measures what it claims to measure [24]. A measure that accurately measures a factor of interest in one culture may not do so in another cultural context [25].

The tools selected for adaptation have been validated in other settings, but not among adolescents in sub-Saharan Africa. They included the Hong Psychological Reactance Scale [26], The Stigma Scale Revised [27] (perceived stigma), the Antiretroviral Medication Attitudes Scale [28] (outcome expectancy), the Consideration of Future Consequences Scale [29], the Child and Adolescent Social Support Scale (CASSS) [30], [31], [32], the HIV Medication Self-efficacy Scale [33], and the World Health Organization's Smoking and Substance Involvement Screening Test (ASSIST) [34], [35]. For family supervision, no existing tool was found that described the types of supervision received by our patients. So, it was decided that for this construct, a simple tool would be created de novo based on adolescents' descriptions of the type of supervision and support they receive when taking medication.

In this study we utilized in-depth interviews with adolescents to establish the construct validity of some Western-developed psychosocial assessment tools while demonstrating limitations in others. We describe the process of culturally-adapting the measurement tools and highlight data justifying decisions to alter aspects of some tools for local use in Botswana. The culturally-adapted tools which required more than minor changes are included in Appendix S1.

Methods

Adolescents were chosen for study enrollment from among the 1472 10–19 year olds who were receiving antiretroviral therapy at the Botswana-Baylor Children's Clinical Centre of Excellence in Gaborone, Botswana. A stratified sampling approach was employed to enroll a minimum of 10 adolescents (5 males and 5 females) in each of 3 age-bands (16–19 years, 13-<16 years, and 10-<13 years). On days when the study was enrolling, adolescents in each enrolling stratum were sequentially approached for enrollment when they presented for routine clinic visits. Enrollment was continued beyond the minimum number of enrollees until thematic saturation was achieved. To be eligible for enrollment, all adolescents had to be aware of their HIV status and had to be prescribed antiretroviral therapy. For>90% of patients in this clinical setting, the mode of HIV acquisition is thought to have been perinatal.

Interviews were conducted in either English or Setswana, whichever language was preferred by the adolescent. The interview guides were translated into Setswana by a single study team member and the translations were subsequently reviewed and improved by all bilingual study team members as well as additional clinic staff. Interviews were conducted by two research team members who were trained to perform qualitative research. Interviews were conducted in a private location within the clinic where the adolescents receive their HIV care. For some of the interview sessions, both interviewers were present. No other observers were present during interviews. Interview sessions were audio-recorded, transcribed and translated into English (if needed). Interviews were transcribed and translated by the same individual who conducted the interview. Following transcription and translation, the audio recordings and interview transcript(s) were reviewed by at least one additional team member to ensure accuracy. The study Principal Investigator subsequently reviewed all transcripts and requested clarifications and alternative translations when aspects of the transcript were unclear.

Interviews were conducted in 3 phases, beginning with the oldest adolescents (age 16–19 years). The interview guide for the oldest adolescents focused on gaining a broad understanding of what the adolescents thought of each of the 5 constructs of interest outlined in Figure 1. It began with broad questions such as “What does stigma mean to you?” Follow-up questions aimed to elicit conversation about how each of the constructs operates in their lives. (e.g., “Some people with HIV fear that people will treat them differently if they know that they have HIV. What are things that people in Botswana sometimes say and do to treat people with HIV differently?”). The concept of psychological reactance was framed for the adolescents using two vignettes, one involving an obedient teenager and the other involving a reactant teenager. In both vignettes, the adolescent is told by her parent that she could not go to a party in the city with new friends. In the obedient vignette, the teen stated that she was not sure that she wanted to go to the party. Since her father did not want her to go, she simply made other plans with different friends. In the reactant vignette, although the adolescent was not sure that she wanted to go the party, she was so angry that her father told her that she could not go that she climbed out of her bedroom window to “sneak out” to the party. The adolescents were asked to describe instances in which they have seen peers respond in ways that similarly demonstrated reactance.

Additional questions were asked to address the appropriateness of the scales chosen to measure social support and substance use. The Child and Adolescent Social Support Scale measures perceived social support from many sources. In order to determine the extent to which measuring these different domains would be important in our cohort, we asked the adolescents to rate on a scale of 1 to 10 how important support was from parents, close friends, teachers, community leaders, church members, and other family members. They were then asked to indicate why they chose their ratings.

Interviews for the middle age group (age 13-<16 years) focused on exploring areas in which the interviews with the oldest group suggested the need for additional clarification. Like the interviews with the oldest adolescents, these utilized primarily open-ended questions and related to the factors indicated in Figure 1. Questions for which thematic saturation was achieved during the interviews with the oldest adolescents were excluded from the interviews with the middle age group. The interviews with the middle adolescents added further exploration of medication-taking autonomy development, parental definitions, and exploring adolescents' goals for the future.

Coding summaries were reviewed by the entire study team between the interviews of the oldest and middle groups and the interviews of the middle and youngest groups. Following the middle age group review, the team adapted the tools based on the adolescents' feedback. Interviews with the youngest adolescents followed the tool adaptation. Because of the developmental limitations of most 10–13 year olds, we used more concrete questioning for this group. The majority of the time spent with the youngest adolescents was spent reviewing each of the survey questions. The interviewer asked the young adolescents to discuss how easy it was to understand each question and whether alternate wording should be considered.

All adolescents in each of the three interview phases were asked to inform the interviewers of anything important to them and their HIV+ peers that was excluded from the interview questions. Each interview took between 30 and 60 minutes. Once transcripts were finalized, they were coded using an analytic framework approach with NVivo9 software [36]. Each script was coded by the study PI and one of two additional team members. We took an analytic framework approach whereby codes were organized into categories relevant to the constructs of interest and additional categories were suggested for themes that arose from the interview data [37]. After each transcript was coded by two or more researchers, discrepancies were reviewed and resolved.

Demographic data were collected and managed using REDCap electronic data capture hosted at Children's Hospital of Philadelphia [38]. Quantitative analyses of demographic data and social support importance rankings were conducted using Stata/IC 11 (STATA Corp. College Station, TX) [39].

Results

The study population consisted of 18 females (53%) and 16 males, with a median time on antiretroviral therapy of 84 months (IQR 60–120 months, range 0.5–216 months). Twenty adolescents (59%) chose to complete the in-depth interviews in English. Of all eligible adolescents approached for study inclusion, only two were excluded (due to lack of parental consent).

Relevance and Understandability of Concepts and Questions

Some Western tools were well-understood and relevant to Batswana youth, requiring only minimal modifications (please see Scale Modifications section below for details). These included the Stigma Scale Revised and the Antiretroviral Medication Attitudes Scale. The ASSIST tool was well-understood with the inclusion of local terms for drugs of abuse. All of the other tools lacked some cultural relevance and/or were in part difficult for Batswana youth to understand. Table 1 presents representative quotations illustrating how the adolescents described the locally-important adherence barriers not represented in the tools. When no interviewer question is listed in the table, the participants provided the information as part of the response to a more exploratory question. Most questions included in the HIV Medication Self-efficacy Scale were relevant to the adolescents. However, absent from the tool were situations the adolescents highlighted as most challenging for maintaining medication adherence. Traveling, particularly school trips, were highlighted. The adolescents also referred to fear of loss of confidentiality when taking antiretrovirals in front of people unaware of their HIV status, such as in the company of friends, visitors in the homes, or visiting relatives' homes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Quotations Illustrating Key Themes.

https://doi.org/10.1371/journal.pone.0109302.t001

Two questions from the Consideration of Future Consequences Scale were considered both relevant and understandable. These included: “I believe that I am in control of what will happen to me in the future” and “I am willing to sacrifice my immediate happiness for things that I expect to happen in the future”. However, the original scale includes abstract concepts about the future that did not resonate with the adolescents despite attempts at explanation. The adolescents expressed concrete concerns about their futures, particularly their ability to marry, have children, obtain employment, support their families, and maintain health as adults.

The Hong Psychological Reactance Scale was not intended to measure psychological reactance related specifically to medication adherence. Yet, the youth noted that adolescents might act rebellious when reminded to take their HIV medications, even if they would not be rebellious in other contexts such as that described for the subject of the vignette. Thus, the concept of reactance was clearly understood. However, the Hong Psychological Reactance Scale also contains jargon such as “acting of my own free will” that was difficult for the adolescents to comprehend.

The adolescents consistently expressed that social support from parental figures is of the greatest importance to them. On a scale of 0–10 with 10 being the most important, 100% indicated that support from parents was 10/10 or “very important”. Most also considered support from other family members to be very important, median (interquartile range (IQR)) score 10 (7–10). In contrast, the median (IQR) score for support from close friends was 5 (5–6). Support from teachers, community leaders and church members were ranked as being of only moderate importance to most of the adolescents with median (IQR) scores of 5 (4–5), 5 (2–5), 4.5 (3–7), respectively. The adolescents referred to a number of relatives as “parents”, with “parental relatives” being those who played the most prominent roles in their upbringing and with whom they felt the closest. “Parents” were described as those who provide for them both emotionally and materially.

Unsolicited Themes: Denial and Ambivalence

Although not solicited directly, the adolescents expressed that issues of denial and ambivalence toward HIV are important to them and their peers. In some subjects, having been without illness for long periods of time seemed to result in a sense of denial about their HIV status. Some noted that medication non-adherence made them feel physically better with the resolution of side effects. This complicated their ability to maintain a sense that the treatment increases their wellness. Some of the most salient quotes related to denial and ambivalence are outlined in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Denial and Ambivalence Statements from Adolescents.

https://doi.org/10.1371/journal.pone.0109302.t002

Discussion

We studied issues of importance regarding adherence to ART among adolescents who have grown up with HIV in a high prevalence setting in order to culturally-adapt relevant assessment tools. We found that some existing tools for measuring psychosocial variables potentially associated with treatment adherence had good construct validity for adolescents in Botswana. Specifically, the Stigma Scale Revised and the Antiretroviral Medication Attitudes Scale were deemed relevant and well-understood. However, other tools that have been previously validated for different populations were found to be lacking for adolescents in our setting. For example, the HIV Medication Self-efficacy Scale failed to include situations in which our adolescents said that their medication self-efficacy is most challenged. The Consideration of Future Consequences Scale did not capture concrete concerns about the future that the adolescents verbalized as relating directly to their ability to maintain their health. The need for medication-specific psychological reactance questions was also highlighted.

Although we know that family plays a more important lifelong role in decision-making in this cultural context than in many Western contexts [40], we were surprised that adolescents in our study ranked the personal importance of peer support well-below that of family and more on par with that of teachers, community leaders and church members. Numerous prior studies stress the importance of peer relationships, support and acceptance to adolescent development and decision-making [41], [42], [43]. Adolescents spend more of their waking hours with peers than with adults [44]. Prior studies in Botswana have confirmed the local importance of peers to some aspects of adolescent decision-making [45]. Since adolescents in our study suggested that peer support is significantly less important than family support, our simplified social support assessment scale allows adolescents to identify a peer, a family member, or any other individual as an important supporter rather than focusing specifically on peer-support.

Because of developmental differences, the same tools may not be valid across the entire age range of 10–19 year olds. One of the strengths of this study was the inclusion of stratified sampling to help ensure that the concepts were both salient and comprehensible across the adolescent developmental spectrum. Other strengths included continuation of interviews until thematic saturation was achieved and the use of a multicultural and multidisciplinary team to analyze and interpret results. The adolescents were sampled from a large clinic which serves children and adolescents from both urban and rural settings. The main limitation of the study is that it did not go beyond cultural adaptation and construct validation to assess other aspects of reliability and validity which require different methods and larger sample sizes. By changing the self-efficacy scale from a 10-point scale to a 5-point scale, we sacrifice granularity and likely will achieve a slightly higher mean score than would be expected with a 10-point scale [46].

These findings stress the need for cultural adaptation of tools that were designed for use in other contexts prior to their use in settings such as Botswana. Unfortunately, existing tools to measure issues of likely importance to adolescent treatment adherence have been developed almost exclusively in low HIV-prevalence settings such as the United States. Cross-cultural adaptation of psychosocial assessment measures is an essential step to ensuring the quality of data obtained using those measures in new cultural contexts [47]. Only when the quality of the cultural adaptation and translation process has been reported can researchers draw appropriate conclusions about constructs assessed and differences between studies and datasets [48]. There is a growing body of literature suggesting that a simple translation and back-translation process is insufficient to ensure the appropriateness of a new test version [49], [50], [51]. Our own prior experience in Botswana suggested that using a multidisciplinary team of skilled bilingual professionals followed by limited pilot testing may fail to produce a culturally-appropriate adaptation of complex tools to measure constructs such as depression and anxiety symptoms [52].

The goal of scale revision is to improve the validity of measures. Next steps for this work will include assessing the internal consistency and test-retest reliability of these tools in a population of HIV-infected adolescents who were not included in this qualitative study. Other techniques such as factor analysis and multidimensional scaling may also be employed to clarify the cross-cultural validity of tools [53], [54]. If deemed to be valid and reliable, the tools will be used to longitudinally assess the relationships between these psychosocial measures and adherence to antiretroviral therapies in adolescents in Botswana. The psychosocial assessment tools developed for use among HIV+ adolescents in Botswana might require minimal adaptation for use in similar settings. Researchers and clinicians seeking to use these tools among HIV-infected adolescents outside of Botswana should consider differences in language and culture as well as HIV prevalence and treatment access in making decisions regarding the need for re-adaptation for their settings. We recommend that researchers seeking to quantify psychosocial constructs among populations in settings where validated locally developed tools do not exist, utilize qualitative methods such as those described in this paper to ensure cultural appropriateness, construct validity and comprehensibility of the measures.

Conclusions

In-depth interviews of HIV-infected adolescents in Botswana confirmed the construct validity of some Western-developed psychosocial assessment tools that could be used to assess the constructs' relationship to treatment adherence in this and potentially similar populations. Denial and ambivalence emerged as underappreciated factors related to HIV medication adherence that need to be explored further.

Supporting Information

Acknowledgments

The study team would like to thank the patients, families and staff at the Botswana-Baylor Children's Clinical Centre of Excellence for making this work possible. We would also like to thank the Botswana Ministry of Health for their commitment to the care and treatment of HIV-infected citizens of Botswana without which many of the adolescents enrolled in this study likely would not have survived childhood. We are also grateful to Frances Barg for her critical review of this manuscript.