Domestic violence among adolescents in HIV prevention research in Tanzania: Participant experiences and measurement issues

Abstract

Objectives

Under-representation of female adolescents in HIV clinical trials may inhibit their access to future prevention technologies. Domestic violence, broadly defined as violence perpetrated by intimate partners and/or family members, may affect trial participation. This study describes violence in the lives of adolescents and young women in Tanzania, explores use of the Women’s Experience with Battering (WEB) Scale to measure battering, and examines the associations between battering and socio-demographic and HIV risk factors.

Methods

Community formative research (CFR) and a mock clinical trial (MCT) were conducted to examine the challenges of recruiting younger (15-17) versus older (18-21) participants into HIV prevention trials. The CFR included qualitative interviews with 23 participants and there were 135 MCT participants. The WEB was administered in both the CFR and MCT.

Results

Nineteen CFR participants experienced physical/sexual violence and 17% scored positive for battering. All married participants reported partner-related domestic violence, and half scored positive for battering. Many believed beatings were normal. None of the single participants scored positive on battering, but one-third reported abuse by relatives. Among MCT participants, 15% scored positive for battering; most perpetrators were relatives. Younger participants were more likely to report battering.

Conclusions

Adolescents experienced high rates of domestic violence and the WEB captured battering from both partners and relatives. The level of familial violence was unexpected and has implications for parental roles in study recruitment. Addressing adolescent abuse in HIV prevention trials and in the general population should be a public health priority.

Introduction

Adolescent girls and young women under age 18 are not routinely recruited into HIV prevention clinical trials despite their risk for HIV (1-3). This under-representation may inhibit their access to future prevention technologies such as microbicides, vaginal rings or pre-exposure prophylaxis if they are not part of efficacy trials. In Africa, where young women shoulder a disproportionate burden of the HIV epidemic, their participation in HIV prevention trials is critical (4). Although there are calls for increased adolescent participation in HIV trials (5,6), a number of factors may affect their HIV risk as well as their ability to join and adhere to trial procedures. Potential barriers to adolescent participation include insufficient understanding of clinical prevention research, issues of parental consent, access to trials and stigma associated with participation (7).

Domestic violence, broadly defined for this study as violence perpetrated by intimate partners and/or family members, is a known HIV risk factor that may also be a barrier to trial participation because it has implications for both recruitment (e.g. formal or informal parental or partner consent) and retention (e.g. financial and social support for continued access to trials) (7-9). Domestic violence includes both physical violence and controlling behavior and it may lead to battering, a particular process which increases women’s vulnerability and powerlessness, thereby potentially affecting her autonomy to choose to participate in an HIV prevention trial (10). In order to better understand the potential influence of domestic violence on adolescents’ trial participation, the first step is to identify an appropriate measure of domestic violence in the lives of adolescents and young women that is informative for clinical trials in Africa.

Partner-related domestic violence is associated with negative reproductive health outcomes including HIV (7,8,11) and may affect research participation if young women do not have partner support. Many women discuss research participation with their husband or partner prior to giving consent and they may be fearful of covert participation if partners are unsupportive (12-14). In Tanzania, intimate partner violence (IPV) is common with 22-29% of women surveyed reporting physical and/or sexual violence by a partner in the last year (11). Among ever-partnered women, 60% experienced two or more controlling behaviors from a partner (e.g. controls her access to health care or insists on knowing where she is at all times). IPV can also start at a young age. Among a sample of students (ages 10-18) in Dar es Salaam, more than a third of girls reported dating violence (15).

However, domestic violence for adolescents in Africa includes more than IPV. Many, if not most, unmarried adolescents and young women live at home with their families and they may be subject to familial violence. Global data on adolescent abuse is scarce, but there is ample evidence that child abuse is common (16). A global survey of children up to age 14 found the highest prevalence of abuse in Africa with 43% of children ever experiencing severe physical abuse and the median percentage of positive attitudes by parents towards corporal punishment was 40% (16). In a nationally representative survey of 13-24 year old females in Tanzania, 58% experienced physical violence by relatives before age 18, primarily by parents, and 52% of those aged 13 to 17 reported physical violence in the past year by a relative, authority figure, or intimate partner (17). In addition, 4% reported being threatened with abandonment by an adult or dating partner prior to age 18. Childhood physical violence was associated with poor health, feelings of anxiety, suicidal thoughts, and sexually transmitted infections in the past year (17).

HIV prevention trials need to recruit young people who are at risk of HIV, but they also want to enroll participants with a high likelihood of being safely recruited and retained in trials. Adolescents and young women who fear violence from disapproving partners or parents may be reluctant to join a trial despite their interest or they may experience difficulty being retained if participation is covert. In order to inform future trials, this study explored violence in Tanzanian adolescents’ lives through qualitative research and a mock clinical trial for a proxy microbide product. In particular, this study examined whether adolescents experienced battering, a potentially more pervasive condition than discrete acts of violence with repercussions for access to healthcare and research participation.

Battering has been defined as “a process whereby one member of an intimate relationship experiences vulnerability, loss of power and control, and entrapment as a consequence of the other member’s exercise of power through the patterned use of physical, sexual, psychological, and/or moral force (pg. 186)” (18). The Women’s Experience with Battering (WEB) scale was developed to capture the psychological trauma and chronic vulnerability women face in violent relationships (10). Although developed for intimate partner violence, we hypothesized that the WEB could also capture the effects of familial violence. Discrete questions limited to physical violence experiences would not necessarily fully capture the vulnerability of adolescents and young women. Information on battering could potentially better inform researchers about the risk context for participants.

Data for this paper come from a larger study conducted in Tanzania to examine the challenges of recruiting and retaining adolescents and young women in HIV prevention clinical trials. The objectives of this paper are:

1. To examine the feasibility of using the Women’s Experience with Battering (WEB) scale with a screener question to assess domestic violence among Tanzanian adolescents and young women ages 15-21; and

2. To examine associations between screening positive for battering on the WEB scale and selected socio-demographic characteristics and HIV risk factors.

METHODS

Data collection included two sequential study phases in Dar es Salaam, Tanzania: community formative research (CFR) and a mock clinical trial (MCT). The CFR was in preparation for the MCT and included repeated in-depth interviews with 23 adolescents (15-17) and young women (18-21). Participants were recruited through clinics, youth centers, schools, NGOs, public areas known as out-of-school youth hang-outs, and study-related community meetings organized with local leaders. Participants were purposefully recruited from low-income neighborhoods and to represent a range of risk contexts (single and married; in- and out-of school; commercial sex workers). All participants were sexually active which was defined as sex within the last three months. Participants were interviewed up to three times in a private locations identified by the participant. Topics included: home life, familial support/conflict; sexual and reproductive health; intimate relationships; HIV risk behaviors; understanding of research concepts; and interest in trial participation. Participants were also administered psychosocial scales for battering (WEB), depression, and self-esteem .

For the MCT, 135 female participants were enrolled and completed the baseline assessment. Eligible participants were aged 15-21, sexually active (as defined above), HIV-negative and not pregnant at baseline. MCT participants were recruited using similar strategies as for the CFR in addition to word of mouth, study brochures, and clientele at the study clinic, the Infectious Diseases Centre (IDC). Interviews were conducted at the IDC, a government clinic that offered youth-friendly reproductive health services. Baseline interviews included socio-demographics, sexual relationship history, partner characteristics, sexual and reproductive health, HIV risk behaviors, and psychosocial scales (e.g. self-esteem, HIV risk perception, WEB).

Ethical approvals were received from Muhimbili University of Health and Allied Sciences, the National Institute for Medical Research, and FHI 360’s Protection of Human Subjects Committee. All participants aged 16-21 provided written informed consent. For those aged 15, parental written consent was required and adolescents provided written assent.

Women’s Experience with Battering (WEB) Scale

The WEB was selected because it captures the chronic vulnerability of women’s experiences with battering as opposed to only the physical markers of violence—battering speaks to the psychological impact of domestic violence (10). We were interested in whether abused adolescents are affected in terms of their HIV risk and their ability to participate in research. The WEB emphasizes the meaning that women attach to the violence they experience and endure in their lives. It has not been used previously in Tanzania, nor any low-income country to our knowledge. Therefore, a thorough process of translation, back-translation, pre-testing, and revision was undertaken.

The WEB is typically administered to women with a sexual partner. However, for our study, not all participants had a regular partner and we wanted to include other non-partner sources of battering (e.g. relatives). We were cognizant of reducing the interview burden for participants; therefore, we administered a screener question to determine who should complete the scale. The following screener question was asked for all CFR and MCT participants: “Thinking about your current situation, is there anybody you live with or spend time with who makes you afraid? This could be a sexual partner, a parent or family member, or a close friend.” If participants answered “yes,” they were asked to give the relationship to this person, then they were administered the WEB. No additional information was asked about the person whom they feared due to ethical issues, but there are no mandatory reporting requirements. However, referral resources were available for participants for additional social and psychological support.

Data management and analysis

CFR interviews were transcribed into Swahili and translated into English. The study team developed a coding scheme and individual raters coded transcripts, checking against the Swahili transcripts as needed. The first few transcripts were coded by multiple raters to check coding agreement and to clarify the coding scheme. Matrices were created to analyze data by participant characteristics and qualitative data were compared against WEB scores. Possible WEB scores ranged from 10-60 with scores of 20 or higher being positive for battering (higher scores indicated higher levels of fear/lack of control; cut-off is international standard set by scale authors) (10). The scale’s internal reliability was good (Cronbach’s α=.84). For the MCT, exploratory association analyses between experiences of battering and selected participant characteristics were conducted using chi-squares or Fisher’s exact tests as appropriate. Those who scored negative for battering included those who did not pass through the screening question (i.e. there was no one they were afraid of).

RESULTS

Community Formative Research

CFR participants experienced high rates of recent domestic violence, including controlling behavior. Nineteen of the 23 participants described experiences of physical violence and/or forced sex and 17% scored positive for current battering (4 of 23). Married women discussed having violent experiences with their husbands while unmarried participants discussed these experiences about both partners and relatives. The four participants who did not discuss any violence experiences were unmarried.

Married women’s experiences of violence

All married participants (n=8) reported partner violence and controlling behavior indicative of battering, but only four had WEB scores positive for battering. From the qualitative data, apparent reasons for the non-battering scores included believing beatings were normal and not reporting being afraid of their partner which was required for administration of the WEB scale.

A pregnant 17 year old married adolescent who was not positive for battering because she did not identify being afraid of anyone, said her husband gets angry when she goes out without permission and he beats her (including while pregnant) if she is not home when he expects her. This adolescent recently got married when she was five months pregnant.

Respondent (R): …with my pregnancy if I go out without telling him, he becomes troublesome and he usually beats me. That is why I don’t like going out without informing him.

Interviewer (I): SO HOW MANY TIMES HAS HE EVER BEATEN YOU?

R: Since we started our relationship, it so many times I can’t even count, because his kind of love is based on fighting; if you wrong/ offend him just a little he will obviously beat you.

I: …SINCE YOU BECAME PREGNANT, HAS HE EVER BEATEN YOU?

R: He has beaten me just twice, but I went to report him to his parent. Now he has stopped beating me, but sometimes when I see that he is upset I get out the room as you know people who are calm/ who don’t speak a lot, they have a beating habit/ anger. So if I see that he is upset, I open my door and I stay out.

A 20 year old married young woman, who also scored negative for battering, said she fears her husband and has received beatings in the past for hanging out with friends whom he considered of bad character. However, she states their quarrels are rare and their conflict is normal, even though it has included forced sex and limitations on her mobility. This young woman got married at age 13 when she became pregnant and now has a seven year old child.

R: If he sees me with other women with bad character, he would warn me not to be friends with her anymore. If I continued being with her, he would beat me and I don’t want my husband to beat me, because he will make me insecure.

I: WHAT HAS HAPPENED TO MAKE HIM BEAT YOU?

R: There was coastal music in a certain place, and he warned me not to go. I asked myself, if my friends have gone, then why shouldn’t I go? I went and after I came back I was beaten. He had warned me about going earlier.

I: HOW MANY TIMES HAS HE BEATEN YOU?

R: Twice

I: HAS IT EVER HAPPENED THAT HE DOES NOT KNOW WHERE YOU ARE?

R: I can say that my husband is very jealous; for instance, when I go places without telling him, he can find out… I must tell him every place I go, where I am going and with whom…I always think that he used to follow me all the time, I do fear a lot.

HOW MUCH CONTROL DO YOU HAVE OVER THE TIMING OF SEX?

R: Control—I don’t have any control, because sometimes I can reject (sex) and he can force me to do (it), sometimes I can reject (him) and he can leave me.

Another pregnant 17 year old married adolescent who did score positive for battering reported being scared of her father growing up as she watched him beat her sister. Currently, she is scared of her husband who beats her if she goes somewhere without telling him.

I: HAVE YOU EVER BEEN AFRAID OF YOUR HUSBAND?

R: I am afraid of him. If I do something wrong, I must be scared of him.

I: WHAT DID YOU DO WRONG THAT MAKES YOU AFRAID OF HIM?

R: For example, if I go somewhere without telling him, I must be afraid, because if I come back he must shout at me or beat me. Or when I borrow things without telling him, a person who lends me that thing might come and demand it in front of him. I am scared of him for that.

Unmarried women’s experiences of violence

Among unmarried participants (N=15), none scored positive on the WEB for battering, but six had ever experienced forced sex by a partner and five mentioned recent physical abuse by relatives. A few mentioned forced sex with current boyfriends but did not discuss being scared of them, nor did they discuss controlling behavior as the married participants did. Physical violence from relatives was usually corporal punishment (i.e. physical discipline) by parents or grandparents; however in one case, a double orphan described abuse from extended family, eventually causing her to leave home. Five participants had previously been or were currently commercial sex workers and one experienced violence from her clients.

A single, out-of-school, 20 year old participant living with her parents, who scored negative for battering, said that her father beats her for coming home late. She also thought he would beat her for other things that might upset him. When asked how she felt about his treatment of her, she responded, “I don’t like it; he is not being fair at all. But, what can I do? He is my father.” She also recalled an event in which she experienced vaginal discharge, thought it might be an STI, and needed money for services.

I: AND WHERE DID YOU GET THE MONEY FOR TRANSPORT OR SERVICES?

R: Since I had no money during that time and that man (her boyfriend) had no money, my mother sent me to the shop to buy something. She gave me money. After some time, I came back and told her that I lost the money. That is how I got the money for transport and hospital services.

I: WHEN YOU TOLD YOUR MOTHER THAT YOU LOST THE MONEY, HOW DID SHE TAKE IT?

R: She beat me. She told me that I was not careful. I let her beat me as much as she wanted, because I really needed that money.

Mock Clinical Trial

A total of 135 MCT participants completed the baseline interview. Eighty-five percent (n=115) of participants were single with regular partners, 7% (n=10) were married or living with a partner, and 7% (n=10) were single with no regular partner. While the CFR was able to purposefully recruit a diverse range of participants, fewer married or partnered adolescents and young women were interested in or able to participate in the MCT. Among adolescents age 15-17, 83% were living with one or both parents compared to 59% among participants age 18-21. Nineteen percent of participants were currently in school, 28% were employed, and 53% were unemployed. A third of participants (33%) had ever been pregnant.

Among the 135 participants, 16% (n=21) stated that they were scared of someone and all but one scored positive for battering based on the WEB (15%, n=20). Table 1 shows the frequency in the report of experience regarding each of the WEB scale items, highlighting that the scale captures the non-physical aspects of battering. People that participants were afraid of included parents (n=6), other relatives (n=12), a sexual partner (n=1) and a friend (n=1), with one respondent not stating whom she feared.

Table 1. Frequencies for agreement with each WEB scale item among all participants who responded “yes” to the screener question: are you afraid of someone? (n=21).

WEB Scale Items	Agree (a little, some or a lot)
	% (n)
1) He/she makes me feel unsafe even in my own home	71 (15)
2) I feel ashamed of the things s/he does to me	52 (11)
3) I don’t do anything to go against his/her wishes because I’m  afraid of him/her	95 (20)
4) When I am in front of him/her, I feel like I react in a fixed way  to a fixed situation	81 (17)
5) I feel like s/he keeps me prisoner	62 (13)
6) He/she makes me feel like I have no control over my life, no  power, no protection	71 (15)
7) I hide the truth from others because I am afraid not to	48 (10)
8) I feel owned and controlled by him/her	71 (15)
9) He/she can scare me without laying a hand on me	86 (18)
10) He/she has a look that goes straight through me	67 (14)

Note: Items 1, 2, 7, 8, 9, and 10 are each missing 1 response but totals are still out of 21.

Table 2 presents the results of the association analysis for the entire sample of 135. Younger participants (age 15-17) and those who were coerced or forced at last sex were more likely to have experienced battering. Among the 41 younger participants (age 15-17), 24% (10) scored positive for battering while among the 94 older participants (age 18-21), 11% (10) scored positive for battering (p =0.039). Forty percent (4 of 10) of those whose last sex was coerced or forced reported experiencing battering (p=0.041). No other associations were found significant.

Table 2. Experiences of battering and characteristics of MCT participants (N=135).

Characteristics	N	Scored Positive on WEB for Current Battering	P-Value
Age
15-17	41	24%	0.039
18-21	94	11%
Living situation
Does not live with either parent	46	22%	0.103
Lives with one or both parents	89	11%
Orphan status
One or both parents have died	52	15%	0.883
Both parents are alive	83	14%
Relationship status
Married or has regular partner	125	14%	0.643
Single	10	20%
Age at sexual debut
16 or younger	50	20%	0.193
17 to 21	85	12%
Last sex coerced or forced
Yes	10	40%	0.041
No	125	13%
Lifetime sexual partners
2 or more	84	15%	0.781
One	51	14%
Ever concurrent partners
Yes	23	17%	0.748
No	112	14%
Ever exchanged money/gifts for sex
Yes	19	16%	1.000
No	116	15%
Ever pregnant
Yes	45	9%	0.171
No	90	18%
Feels at risk of HIV a little or a lot
Yes	105	15%	1.000
No	29	14%
Condom use at last sex to prevent HIV/STIs
Yes	65	15%	0.857
No	70	14%
Family planning (FP), excluding condoms
Never used modern FP method	114	16%	0.738
Ever used modern FP method	21	10%
Depression
Moderate or severe	37	22%	0.171
None or mild	98	12%
Self-esteem
Low	12	8%	1.000
Normal or high	123	15%

DISCUSSION

There are numerous ethical and logistical challenges for adolescent participation in HIV clinical trials (19). In contexts where partner or parental consent and support is explicitly or implicitly required for trial participation, domestic violence, and more specifically battering due to its psychological impact, may affect study recruitment and retention. This study examined whether the WEB could be an informative scale for clinical trials—researchers could use the scale to screen for battering among potential participants and/or use the scale to help identify factors related to trial retention. This study documents that adolescents experience not only domestic violence, but battering more specifically.

The WEB scale had good internal consistency and it detected battering from both partners and relatives. It did not capture all aspects of domestic violence, nor should it have. In particular, among married participants, use of the screener question missed some of the impact of IPV largely due to beliefs that violence and controlling behavior are normal or to not indicating that they were afraid of their partner. Another Tanzanian study similarly found that women exposed to IPV do not necessarily report being fearful of their partner (20). The WEB also captured familial battering. However, the qualitative data illustrate that not all familial violence translates to battering nor will it necessarily be detected by the WEB, particularly if participants do not identify a primary person of whom they are afraid.

Assessing exposure to violence and battering is important for the health and safety of HIV trial participants. For researchers that may utilize the WEB in future trials, we recommend administering it to all participants with regular sexual partners as the WEB was intended as well as single participants who can identify a primary person about whom to answer the questions. The screener question could be revised as “is there anybody you live with or spend time with who makes you afraid or who has been (physically and/or sexually) violent towards you?” Because the qualitative data revealed that some participants were not afraid of the persons committing violence, this broader revised question may possibly screen in more participants.

The level of familial battering in the MCT was unexpected and has implications for parental roles in research. Although studies are ethically subject to strict confidentially protocols, if a participant is experiencing battering and/or participating in a study covertly (assuming she can consent herself), researchers should have additional safety measures such as alternative communication for follow-up (e.g. using a friend’s phone number and not the family’s, or partner’s, shared phone). Besides being cognizant of confidentiality issues, researchers should be knowledgeable on the local laws and social services available to address adolescent abuse.

There were some study limitations. Because the WEB was not administered to every partnered participant, we are unable to assess the overall level of battering in intimate partnerships. We also did not assess if participants experienced violence and potential battering from multiple sources. For the MCT which required repeat attendance at a clinical setting, there is potential selection bias for those who chose not to participate due to fear of violence. Another limitation is that the statistical analysis for the MCT data should be interpreted with caution since the sample was purposively selected. Generalizability is limited although we hope this self-selected sample of participants, recruited in the community similar to other trials, represents the type of participants who enroll in microbicide clinical trials. In addition, the sample size for some of the groups may be too small for conducting an association analysis with adequate power. With these caveats in mind, we found a higher prevalence of battering among the younger group that might be worthwhile exploring further.

Domestic violence is associated with risk of HIV and battering in particular may affect one’s ability to autonomously choose to participate in research. Globally, adolescent domestic violence is not on the public health radar. Greater public awareness along with better resourced research and programs needs to be paid to strategies for prevention of adolescent abuse from relatives as well as partners in low-resource settings in the context of HIV trials and the general population.