RESEARCH ARTICLE Open Access

A reflection on ethical and methodologicalchallenges of using separate interviewswith adolescent-older carer dyads in ruralSouth AfricaDumile Gumede1,2* , Nothando B. Ngwenya1, Stella Namukwaya3, Sarah Bernays4,5 and Janet Seeley1,3,5

Abstract

Background: This article discusses our reflections on ethical and methodological challenges when conductingseparate interviews with individuals in dyads in the uMkhanyakude district, South Africa. Our work is embedded in anethnographic study exploring care relationships between adolescents and their older carers in the context of a large-donor funded HIV programme. We use these reflections to discuss some of the challenges and present possiblemanagement strategies that may be adopted in conducting dyadic health research in resource-poor settings.

Methods: Drawing from the relational agency, three rounds of separate interviews and participant observation wereundertaken with dyads of adolescents aged between 13 and 19 and their older carers aged 50+ from October 2017 toSeptember 2018. A reflexive journal was kept to record the interviewer's experiences of the whole research process. Weidentified methodological and ethical challenges from these data during the thematic analysis.

Results: A total of 36 separate interviews were conducted with six pairs of adolescent-older carer dyads (n = 12participants). Five themes emerged: recruitment of dyads, consenting dyads, confidentiality, conducting separateinterviews with adolescents and older carers, and interviewer-dyad interaction. We also illustrated how we dealt withthese challenges.

Conclusions: Results from this study can guide the recruitment, consenting and collecting data for health studies thatemploy a similar form of enquiry in LMICs. However, ethical and methodological challenges should be recognised asfeatures of the relationships between cross-generation dyads rather than weaknesses of the method.

Keywords: Dyads, Adolescents, Older carers, Confidentiality, Ethics, Separate interviews

BackgroundQualitative studies on dyads have increased [1] since thedyadic research approach emerged in the 1970s in mar-riage and family studies [2]. The approach examines ‘thedyadic perception of reality, dyadic meaning, and dyadicbeing-in-the-world, in addition to the relationship com-ponent’ [1]. For some research questions, interviewingdyads may generate data that could not be obtained

from interviews with individuals [2]. Scholars have usedthe dyadic approach in studies involving couples [2],caregivers and children or grandchildren [3–6] and pa-tients and carers [7–10] in health research. For example,a Ugandan study examined care dyads of caregivers andHIV-positive young people about their experiences ofdisclosure of HIV status and the influence it had on theirrelationship [5]. Different interviewing techniques havebeen used to collect data with dyads. The interviews canbe conducted separately with each dyad member [11] orjointly with both members together [12, 13]. Separateinterviews enable each individual to respond from theirown perspective, ‘capturing the individual within the dyad,without forgoing the dyadic perspective’ [1], whereas

© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

* Correspondence: dumile.gumede@ahri.org; Gumeded@ukzn.ac.za;dumileg@gmail.com1Africa Health Research Institute, Durban, South Africa2School of Applied Human Sciences, University of KwaZulu-Natal, Durban,South AfricaFull list of author information is available at the end of the article

Gumede et al. BMC Medical Ethics (2019) 20:47 https://doi.org/10.1186/s12910-019-0383-9

conducting interviews jointly results in a shared narrative[14]. A separate interviewing approach raises methodo-logical challenges related to the method of conductinginterviews with each individual member of a dyad and itsinfluence on the data collected [14]. Separate interviewscan be time-consuming for the dyad members andthe researcher because two interviews are carried out[15]. It also raises ethical concerns related to the per-sonal relationship between dyad members when theyare interviewed separately [16]. Therefore, conductingseparate interviews with dyad members could posepotential methodological and ethical problems thatshould not be overlooked.Methodology and ethics are interrelated. Adherence to

ethical standards can add to the value of research and,conversely, methodological soundness can strengthenethics [17]. In their editorial article, Knottnerus and col-leagues [18] explain the link between methodology andethics that ‘ethics of research methodology requires amethodology of research ethics’. Reviewing the ethicsof research should consider whether the researchquestions are worth asking and if the methods usedare an effective way of answering them [17]. The ap-proach of conducting separate interviews as a method-ology has ethical implications as it critically influencesdata collection, interpretations of data and reportingof data.Ethical issues arise in all aspects of research and are

particularly salient when studying vulnerable populationssuch as children and adolescents, older people andpeople living with HIV or affected by HIV [19–21].Researchers have ethical responsibility for ensuring thatindividuals are given all the information needed to makeinformed decisions about whether to participate in re-search or not. The process of respecting autonomy iscomplex, given different conceptions of personhood[22]. Most persons define themselves and make deci-sions within a wider network of social relationships [22,23]. When engaging with an individual, one is, in fact,dealing with a complex relational web of persons whomay include the immediate family, peers and significantothers for whom the decision matters [22]. However, torespect autonomy, Osamor and Grady [23] suggest re-searchers need to understand and respect relationshipsthat are important to individuals and the process withwhich they incorporate the values inherent in these rela-tionships into their decision-making. Adolescents mayagree to participate out of obedience to or respect fortheir caregivers [19, 20]. For example, in a study tounderstand the resolution of discordance betweenadolescent-parent dyads about participation in research,Francis and colleagues [24] report that sometimes oneindividual in the dyad asserted authority over his/herpartner about the decision for participation.

Confidentiality is one of the cornerstones of researchinvolving human participants. Protecting research partic-ipants’ right to confidentiality is a responsibility sharedby researchers, institutional review boards, and partici-pants themselves. However, combining ethnographic anddyadic approaches adds further complications to main-taining the confidentiality of individuals within the samedyad. It is difficult to maintain confidentiality betweenthe members of the dyad when information is validated,or when different versions are compared [1]. The chal-lenge of confidentiality also arises when reporting findingson the dyad as the amount of information that could iden-tify dyad members could be more than in individual inter-views [16]. Separate interviewing has the potential togenerate anxiety within dyads because this approachmight suggest that secrets exist, and that one person iswilling to share these secrets with the researcher (and notwith his or her partner in the dyad) [14]. Individuals mayexpress the desire to know what the other person in thedyad said, placing the researcher in an awkward position[25, 26]. The method of separate interviewing can con-front the researcher with a dilemma of how to make senseof different versions of a story regarding events in individ-uals’ lives [11]. Furthermore, in the face of competing ac-counts from individual interviews, the researchers arelimited in their ability to probe further and ask directquestions, since in so doing, they may unintentionallydisclose what the other individual said, therebybreaching confidentiality [11, 25, 27].There may be an expectation of short and long-term

benefits and advantages to participation that influencesindividual consent and participation. This challenge maybe greater when doing ethnographic work among vul-nerable populations, because of the extended amount oftime spent with participants [20]. In a project with chil-dren and women affected by HIV/AIDS in Kenya,Nyambedha [20] found that the people in the area didnot differentiate between activities of non-governmentorganizations (NGOs), other researchers and his ownstudy. Consequently, there were raised expectations thatparticipation in the research could lead to interventionsthat would assist participants. Nyambedha argues thatresearchers can cause harm if no action is taken to ad-dress the high expectations participants may have.As shown, researchers face complex questions of

methodology and ethics in the application of a separateinterviewing approach. These challenges are of a com-plex nature, yet dyad studies have received a relativelysmall amount of critical ethical and methodological at-tention [12, 16, 27, 28], particularly in low-middle-income countries (LMICs).In this article, we draw upon our experience of con-

ducting separate interviews in an ethnographic projectwith care dyads of adolescents and their older carers in

Gumede et al. BMC Medical Ethics (2019) 20:47 Page 2 of 12

rural South Africa and examine the ethical and meth-odological challenges, which have arisen from our workand how these were addressed. We use the reflections todiscuss some of the challenges and present possiblemanagement strategies that may be adopted in conduct-ing dyadic health research in LMICs.

Research design, method and sampleThe study employed an ethnographic approach to ex-plore care relationships between adolescents and theirolder carers in South Africa, and the contextual factorsassociated with HIV risk among these adolescents. His-torically, the care of orphaned and vulnerable childrenand adolescents in Africa is often provided by oldercarers in extended families [29]. The onset of the HIVepidemic left many adolescents orphaned, vulnerableand in need of care [29, 30]. In many cases, those whofill the care deficit are older women [30]. The increasedcaring responsibilities by older carers commonly coin-cides with a time when many older people themselvesneed care. Older carers may experience health problemsdue to age [31, 32] and caregiver stress [29], which im-pacts negatively on the care relationship and the kind ofcare given [31]. It is important to understand howadolescent care is provided by older carers and expe-rienced to improve the kind of care that adolescentsreceive, and indeed, the reciprocal care the older per-son may benefit from.The theoretical framework of relational agency in-

formed the research question, methods and interpret-ation of findings. Relational agency demonstrates thatindividuals’ experiences are shaped and influenced bytheir relationships with others [33]. Relational agency re-fers to how the ability of someone to exercise agency isbounded by their relationships, as well as the structuresof their environment. Their relationships, which are dy-namic and evolving, are crucial structures within theirenvironment. In adolescence, it may be a particularlyuseful theoretical construct because of the fluidity andrapidly changing nature of young people’s relationships.The influential relationships, which are guiding their be-haviour, are changing too as social and peer relationshipsand influences proliferate.The increasing attention paid to relational agency in

shaping adolescent’ experiences and health-seeking be-haviour shows there is clearly a need for more dyadicresearch, which will provide insights into these relationalexperiences, challenges and solutions that should dir-ectly benefit the provision of effective support for bothadolescents and their older carers in LMICs. In additionto capturing the young people’s perspective, understand-ing how the relationship between the older carer andyoung person is experienced and how this affects careand well-being is important. The dyadic approach to

interviewing offers considerable potential to explore howrelationships of individuals in a dyad influence data col-lection and ethics processes.The study was conducted at the Africa Health Re-

search Institute (AHRI), in the uMkhanyakude district ofnorthern KwaZulu-Natal, South Africa, in the context ofa large donor-funded programme in KwaZulu-Natal,South Africa. Different implementing organisations de-livered the programme activities in the uMkhanyakudedistrict to address HIV risk behaviours, HIV transmis-sion, and gender-based violence.We purposively sampled adolescent-older carer dyads

to achieve an in-depth understanding of the participants’experiences. Participants were selected according to thefollowing criteria:

� An adolescent girl or boy aged between 13 and 19years living and cared for by an older carer (eitherolder man or older woman) aged 50+ years in theuMkhanyakude district.

� An adolescent who was the recipient of at least oneHIV behavioural intervention from the selectedimplementing organisation.

� An older carer who was the primary caregiver for anadolescent girl or boy who was eligible based on theabove criteria.

An implementing organisation provided permission toconduct the study and to recruit its recipients throughprogramme facilitators. The programme facilitators ap-plied the inclusion criteria to identify dyads to be invitedto participate in the study.The first author, a woman with experience in qualitative

interviewing, conducted participant observation and threeseparate interviews with six adolescents (aged 13–19) andsix older carers (aged 50+) from October 2017 to Septem-ber 2018. In total, thirty-six interviews were conducted. Weadopted a separate interviewing approach with each partici-pant being interviewed one on one. Conducting the inter-view in private was a priority. Even if the other individual inthe dyad was also at home, the interview was conducted ina separate space without anyone else being able to overhear.The interviewer used a similar interview guide with eachperson to explore issues about the participants’ back-ground, relationship with older carer/adolescent, parentingpractices, communication about sexual and reproductivehealth, and experience with interventions. However, weadjusted questions to be more appropriate for differentages. The interviewer also visited dyads at home to conductparticipant observation at different times of the day. Thesedata were recorded in field notes focusing on what was ob-served, informal conversations with participants, records ofactivities, conversations between participants, and partici-pants’ nonverbal and verbal behaviours.

Gumede et al. BMC Medical Ethics (2019) 20:47 Page 3 of 12

The first author kept a reflexive journal in which shenoted her experiences of the whole research process.The journal contained concrete descriptions of the inter-viewer’s experiences about ethical and methodologicalchallenges emerging in this dyad study of care relation-ships between adolescents and their older carers andhow these were addressed.Interviews were audio-recorded, transcribed and trans-

lated into English. Transcripts and field notes werecoded and managed using NVivo 11. Following dyadicanalysis, we examined themes emerging from each dyadmember’s narratives [1] by assessing contrasts and over-laps between the individual versions. This paper doesnot aim to report on the substantive findings of this ana-lysis, but rather we focus on ethical dilemmas emergingin this dyadic study of care relationships between adoles-cents and their older carers. We identified methodo-logical and ethical challenges from these data during thethematic analysis.

Ethical considerationsThe study received approval from the University ofKwaZulu-Natal Humanities and Social Sciences Re-search Ethics Committee (ref HSS/1109/017D). We ob-tained written informed consent from all participantsbefore they took part in the study. For adolescents agedless than 18 years, we obtained written consent fromtheir older carers and written assent from the young per-son themselves. No names of participants were recorded;instead, participant codes were used, and the names oforganisations anonymised.

ResultsParticipant profileSix adolescent-older carer dyads (n = 12 participants)took part in this study. All the six older carers werewomen aged between 56 and 80 years and caring for be-tween two to 15 grandchildren. Two were paternalgrandmothers, and four were maternal grandmothers.This reflects the gendered nature of caregiving commonin this setting. Some older carers started caring forgrandchildren since childbirth, while others assumedcaregiving responsibilities when grandchildren in-migrated to the older carer’s households. The number ofyears as carers ranged between three and 14 years. Allolder carers relied on either a state pension or child sup-port grant for a source of income. Of the six oldercarers, only one was cohabiting with a sexual partner;while others were widowed. Only two older carershad attended primary school. Although not ques-tioned directly about their HIV status, two oldercarers disclosed that they were living with HIV andon ART.

There were five adolescent girls and one adolescentboy in the study, between the ages of 13 and 19 years.Four were still in secondary school while the other twohad dropped out of school due to pregnancy and aca-demic difficulties. Of the six adolescents, all were recipi-ents of either one or two HIV prevention programmesfrom a single implementing organisation. All adoles-cents’ biological mothers were still alive, and four werepaternal orphans. There were various reasons why thebiological mothers were not providing primary care tothe adolescents including maternal abandonment, non-marital births, unemployment and migration.

Ethical and methodological challengesThe information recorded in the first author’s journalhighlighted several ethical and methodological chal-lenges, which guided the choice of the following themes:recruitment of dyads, consenting dyads, confidentiality,conducting separate interviews with adolescents andolder carers, and interviewer-dyad interaction, and howthese challenges were addressed in the study. It must benoted that some challenges were a combination of boththe method and ethics. We have thus indicated whethereach challenge is a methodological, ethical or mixedchallenge.

Recruitment of dyadsPotential sampling biasWe anticipated that recruiting participants through theimplementing organisation had potential sampling biasprior to the recruitment process. The programme facili-tators relied on their self-knowledge to identify potentialparticipants, as the facilitators were local communitymembers themselves. Although this was an effective wayof identifying participants, there was potential for biasabout individuals that the programme facilitators choseto approach. Given that the study involved exploringparticipants’ experiences with HIV behavioural interven-tions, programme facilitators might have been temptedto lean towards inviting people who were thought likelyto provide a more positive account of the organisationand the interventions than others. To reduce potentialsampling biases, we worked with two different programmefacilitators for each to identify two pairs of adolescentsand their older carers. We also integrated a participant-driven sampling technique by requesting two adolescentparticipants to refer the researcher to other adolescent re-cipients in the care of older people. Having a small sampleof participants who know each other in the neighbour-hood raised further concerns about maintaining the confi-dentiality of results. To keep the dyad partners’ versionsconfidential from each other, we only checked our inter-pretations with individuals, treating them as separatesources. To enhance the anonymity and confidentiality of

Gumede et al. BMC Medical Ethics (2019) 20:47 Page 4 of 12

individuals, we needed to omit or slightly alter some dis-tinguishing information, which could identify individualsfrom another [1]. We did data omissions and alterationsto balance the need to preserve both contextual issues andconfidentiality [1]. For example, we either omitted or al-tered information such as names of birthplaces, names ofschools and church names. In addition, we provided allparticipants with a pseudonym to ensure confidentialityand to protect the privacy of the individuals.

Risk of coercion to participateAn ethical challenge that emerged from the recruitmentof dyads was the risk of coercion to participate. Firstly,we suspected that there was potential coercion during therecruitment as some individuals may have felt under someobligation to participate if approached by programme fa-cilitators. It was even more concerning when the re-searchers learnt that the facilitators were local communitymembers, having established relationships with potentialstudy participants in the area. For example, one of the fa-cilitators was a pastor’s wife and running a local crèche. Itwas highly probable that some individuals were membersof the same church with the facilitator. We were con-cerned that some individuals were likely to feel obliged toparticipate in view of maintaining the existing relationshipwith the facilitator by not refusing to participate. Secondly,older carers knew their adolescent grandchildren partici-pated in HIV behavioural interventions. Approachingolder carers first and for them to provide initial agreementmay have had implications for how ‘free’ adolescents wereto then refuse to participate. Similarly, approaching ado-lescents first could have been perceived as disrespectful tothe older carers. We were concerned about the potentialof members influencing or coercing each other for partici-pation in the study. An option for approaching them to-gether was difficult as it was not easy to find them both athome at the same time. Older carers were more accessibleto contact and meet, given that they were mostly at home,as they were unemployed and spent their time doinghousehold activities. Whereas, adolescents were more dif-ficult to contact as they were frequently away from home.All potential approaches involved compromises and risks.However, the implementing organisation guided us as tothe most appropriate strategy to reach the participants. Inall cases, the older person was contacted first and then ef-forts made, on approaching the young person, to makethem at ease and able to make their own decision aboutwhether to take part. The programme facilitator and theinterviewer visited the individuals together at home for anintroductory meeting; thereafter the interviewer was leftwith the individuals to discuss the study and to seek con-sent for the older carer and their adolescent grandchild.The interviewer emphasised voluntary participation to theadolescents in view of reducing the risk of coercion by the

older carers and the programme facilitators. All the sixdyads approached, consented to participate in the study.There were no refusals to participate or withdrawals fromthe study.

Non-contactable dyad individualsRecruiting dyad individuals was time-consuming giventhat adolescents were often away from home, so repeatvisits were required. In addition to school-related activ-ities, young people were involved in doing householdchores outside of the home such as fetching wood andwater, attending church-activities or visiting family rela-tives. Both members had to be first contacted and in-vited for participation prior to consenting andinterviewing. Failure to contact the other member haltedthe process until both consented as we did not know ifthe other member would agree to participate or not.Commonly, the older carer assisted the researcher withinformation about when the adolescent was likely to behome and contactable. This required repeat visits untilcontact was made with an adolescent. Nonetheless, theinterviewer established a stronger rapport with oldercarers during those repeat visits, which somehow influ-enced unintended preferential treatment given to someparticipants by the interviewer. This will be explainedfurther in the section discussing the challenges ofinterviewer-dyad interaction.

Consenting dyadsDiffering needs of dyad membersThe process of consenting dyads was both ethically andmethodologically complex. Differing needs of individualsin dyads challenged the provision for informed consentand assent during the process. Some participants consid-ered the study information sheet to be too long. All theolder carers wanted the sheet read to them because theycould not read themselves due to literacy skills or visionproblems. Some non-literate older carers were not com-fortable signing the consent form. In contrast, the ado-lescents did not want to spend the time reading theinformation sheet through and preferred the researcherexplain it to them. They wanted the consent process tobe accelerated to fit in with their limited time that theywere willing to devote to the research given their com-peting commitments. As a result, we adapted the in-formed consent and assent process to the needs ofindividual dyad members such as reading or paraphras-ing the sheet and the use of a mark in place of a signa-ture on the consent document. However, rushing theprocess of informed consent raised the dilemma of beinguncertain about participants’ full understanding of in-formed consent. We ensured that the interviewer presentedthe study information to the study participants at every

Gumede et al. BMC Medical Ethics (2019) 20:47 Page 5 of 12

repeat interview and confirmed with the participants if theywere still willing to continue participation in the study.

Parenting styleWe observed different parenting styles among oldercarers during the process of informed consent. Someolder carers demanded total control of the process andexpected children to obey their decision about researchparticipation, while others encouraged adolescent auton-omy in research participation. For example, B1 (femaleolder carer aged 64) consented to participate in thestudy and the interviewer informed her that B2 (adoles-cent girl aged 14) was also required to decide on herown about participation in the study. In her response,B1 told the interviewer not to bother because she wouldinstruct B2 to participate. Knowing that valid informedconsent was essential to enable individuals to be fullyaware of what they were taking part in, the interviewertook time to explain the principle of voluntary participa-tion until the older carer understood and linked this totheir knowledge of the political changes in South Africaabout children’s rights. In contrast, older carer E1 (fe-male older carer aged 56) expressed that she was happyfor E2 (adolescent girl aged 13) to make her own deci-sion to participate or not because, although regarded achild, she was capable of doing so.

ConfidentialityParticipants’ limited trust in the confidentiality processSome participants displayed limited trust in the confi-dentiality process, concerned that information might getto the other dyad member or even if it did not, therewere concerns about being seen to voice criticism aboutthe other member of the dyad. To illustrate this, wedraw two examples from the field notes. In the first ex-ample, an adolescent girl was uncertain about disclosingconfidential information to a stranger, hence the adoles-cent wanted assurance that the interviewer would notshare information with her grandmother:

“Today, I had the first interview with C2 (adolescentgirl aged 15) who lived in the care of her 80-year oldgrandmother (C1). Prior to this interview, I hadalready interviewed C1 on the previous day, whereshe narrated her perspective of her relationship withC2. Amongst her concerns, C1 suspected that C2 waspregnant. Therefore, I went to interview C2 havingthis confidential information that her grandmotherand others in the family were suspecting that C2 waspregnant. During the interview, while I was keen tofind out from C2 about the suspicions of her preg-nancy but I did not ask her about it. She also did nottalk about it until towards the end of the interview,when I asked her if she had any other thing she

wanted to tell me before we closed the interview. Shestarted mumbling, facing down and hesitant to talk. Imotivated her to feel free to talk to me. She lookedstraight into my eyes, and asked, ‘are you going to tell mygrandmother that I am pregnant?’ This was her secret, asshe was not planning to tell her grandmother. Shepreferred her grandmother to discover the pregnancy onher own; otherwise, she was nervous she could beexpelled at home if her grandmother discovered thepregnancy in its first trimester. As a result, she wantedassurance and commitment from me that her secretwould not be passed on to her grandmother”.

In the second example, an older carer was hesitant tonarrate confidential information about her grandson andhis mother because she feared that her grandson mightlearn if the interviewer did not keep the informationconfidential:

“My second interview with A1 (older carer womanaged 64) revealed that A2 (adolescent boy aged 15)was not happy about using his maternal surnamehence he was in a process of changing to his paternalsurname. When trying to understand the reasons forchanging the surname, before she could answer A2raised up her hands (as if surrendering) and said: ‘it’sdifficult to explain people’s secrets [grandson and hismother] … please protect me by keeping this as a secret’.She wanted a guarantee that her grandson would notfind out that she divulged his secrets to me”.

Both A1 and C2 had limited trust in the confidentialityprocess and needed assurance. They were concernedthat breach of confidentiality would result in a conflictwithin the relationship.To enhance individuals’ trust and confidence in the con-

fidentiality process, we employed several strategies. Firstly,at every start and end of interviews, the interviewer con-sistently assured each individual that the researcher wouldnot disclose information to the other dyad member. Sec-ondly, we also secured individuals’ confidence through theinterviewer’s conduct of adhering to the principle of re-spect for individual confidentiality by not disclosing infor-mation between dyads. Lastly, the nature of repeatinterviews allowed the trust to develop which had beenearned through their experience of the first interview.

Fear of unintended disclosure of information by theinterviewerThere was no accidental disclosure of information bythe interviewer to dyad members; however, there wasfear for the interviewer to mix up individual stories be-tween adolescents and their older carers. Mixing up in-formation could have led to unintended disclosure of

Gumede et al. BMC Medical Ethics (2019) 20:47 Page 6 of 12

information. To avoid any unintended disclosure, beforeconducting subsequent interviews in the same dyad, theinterviewer carefully read and reread the summary andlistened to the audio of the member’s previous interviewto recollect interview content with that member. Theinterviewer also took rigorous reflexive journal notesthroughout the data collection process to maximiseawareness of potential threats to confidentiality and con-tinuously reflected on best practice to protect eachmember’s confidentiality. This enhanced adherence toconfidentiality by guarding against slippages.

Right to access confidential information of the other dyadmemberIt was not only about the researcher not disclosing infor-mation unintentionally but also managing requests forbreaching confidentiality by a participant. Some partici-pants, particularly older carers, expected the interviewerto inform them about issues discussed with adolescents.For example, after interviewing C2 (adolescent girl aged15) at home, C1 (older carer aged 80) immediatelyapproached the interviewer to demand informationabout her adolescent granddaughter from the inter-viewer. She was suspicious and wanted to find out fromthe interviewer if the adolescent was indeed pregnant.The older carer wanted to overrule confidentiality andbe given the right to access information because of herexpectation of authority over the adolescent. This com-plicated the dyadic research because the interviewer wastrying to retain their respect and trust while disagreeingwith and not complying with the carer’s request for in-formation. The interviewer politely reminded partici-pants about the confidentiality clause and affirmed thatby so doing she wanted all participants to trust herabout their information. This approach appeared effect-ive because attempts to seek information about the othermember stopped.

Conducting separate interviews with adolescents andolder carersExpectations of benefits for research participationSpending extended periods of time with the same groupof participants generated expectations of benefits amongthe participants for their participation in the study. Asthe participants were narrating their problems to theinterviewer, they hoped for solutions from the inter-viewer. It was the older carers who had raised expecta-tions of assistance, although one adolescent girl (C2)who was requesting for a loan on behalf of her grand-mother (C1) to settle a financial debt as their lives wereunder threat. The older carers expected assistance withemployment opportunities, food, housing, medicationfor arthritis and minor ailments, counselling, money,and adult schooling opportunities. It was interesting to

note that the expectations of benefits were required toaddress their relational needs. For example, older carersrepeatedly reported the challenges of caring for theiradolescent grandchildren and their concerns about theadolescents’ risky behaviours. Participants mentionedpoor communication between the older carers and theadolescents as a challenge in the relationship. The oldercarers often requested the interviewer to intervene instrained dyad relationships and expected the interviewer toprovide some form of counselling. Being caught in the mid-dle of strained relationships between dyad members wasstressful for the interviewer, as she did not anticipate themagnitude of relationship problems yet could not help.Several measures were taken to address these con-

cerns. First, all the participants were offered snack packsas a token of appreciation at every visit. The participantsappreciated the packs, as they would share the items inthe family. Second, the interviewer continued to explainher role as a researcher, not a therapist, in accordancewith the study protocol. However, to compensate for theresearcher’s inability to help, participants were providedwith a self-referral list with contact details of local serviceproviders, which they could contact for support, and theresponsibility to contact the service providers was left withthe study participants. This raised an ethical concern thatthe information support that we provided for self-referralswas very difficult for our participants to then take forwardin this setting. For many study participants, a lack of fi-nancial resources was a barrier for them to access servicesdue to related costs such as transport. Mostly older carerswere non-literate and could not make phone calls, andothers did not have access to telephones to contact serviceproviders. Participants could not understand why theinterviewer did not contact service providers on their be-half, as they did not have resources to do so. We wereconcerned that participants were likely to lose interest inthe study if they thought researchers were not supportive.Ideally, facilitating the linkage between study participantsand the service providers would have been the best so-lution; but this had budgetary implications, whichwere not covered in the study. Our strategy was tospend more time explaining the role of research,thereby emphasising its remit. This approach appearsto have ameliorated misunderstandings, but it illus-trates the challenges of conducting such relationallyfocused research in low-income settings.

Contradictory storiesParticipants’ accounts revealed some contradictionsabout a shared experience between individuals in a dyad.Contradictory stories came up on issues where onemember said the opposite of the other member about anevent. Below is an extract of field notes illustrating a

Gumede et al. BMC Medical Ethics (2019) 20:47 Page 7 of 12

contradictory story between an adolescent girl and herolder carer:

“Today, I interviewed both B1 (older carer womanaged 64) and B2 (adolescent girl aged 14) separatelyfor the first time. B2 was interviewed first. Whenasked about her experiences in the HIV preventionprogramme for adolescents and caregivers, B2revealed that neither her grandmother (B1) noranyone from the family attended the programmesessions with her. According to B2, she attended theprogramme alone without her caregiver. To ourknowledge, the programme was designed foradolescents and their caregivers. Similarly, in theinterview with B1, she also stated that she did notparticipate in the programme with B2. However, B1differed from B2 in that B2 attended the programmesessions with her aunt [B1’s youngest daughter]”.

Probing contradictory stories was difficult as we werecareful about asking direct questions, which could revealthe other dyad story, thereby breaching confidentiality.It was also not possible for the members to contest theircontradicting stories as they were interviewed separately.This contradiction was not resolved; we accepted thateach dyad member had a different version. Conformingto dyadic research, our focus was to look at perspectiveson the same relational experience.

Single-sided account of a storyMethodological and ethical dilemmas were intertwinedwhen a single-sided account of a story was experienced.This was when a dyad member presented information,which the other dyad member was discreet about. Forinstance, F1 (older carer aged 76) and E1 (older careraged 56) were interviewed before their adolescent grand-children. During the interviews, they both disclosed thatthey were living with HIV and on ART. The older carersshared that they had disclosed their HIV status to theiradolescent grandchildren and that the adolescent grand-children were supportive of the grandmothers to adhereto treatment. Interestingly, both the adolescent grand-children F2 (adolescent girl aged 19) and E2 (adolescentgirl aged 13) made no reference to their older carers’HIV status during the interviews. The interviewer wascaught in a dilemma about the extent of probing the sec-ond dyad member (adolescents) about the accounts ofthe first dyad member (older carers) while being mindfulabout the protection of confidentiality. Instead of tryingto have adolescents’ talking about their older carer’s HIVstatus, we recognized that individuals had their story totell in their own way. This also illustrated the nature ofthe care relationship.

Challenges of interviewer-dyad interactionImpossibility of matching age and gender characteristics ofinterviewer with dyadsAll interviews were conducted in isiZulu by the first au-thor, a middle-aged woman. The age and gender ofinterviewer did not match equally with all individualdyad participants. All the six older carers were womenbut older than the interviewer. Because of age differ-ences, older carers viewed the interviewer as theirdaughter by addressing her as ‘my child’ or ‘my daugh-ter’. On the contrary, the adolescents were younger thanthe interviewer hence they perceived the interviewer as‘a mother’ or ‘an aunt’. These age and gender differencesinevitably influenced the interaction between the partici-pants and the interviewer. On the one hand, the age ofinterviewer facilitated and strengthened the relationshipbetween the interviewer and older carers. On the other,during the first interviews, it was not easy for adoles-cents to explicitly express their experiences to an olderinterviewer whom they regarded as a mother. Adoles-cents tended to pause more often and for long periodsand avoided eye contact with the interviewer. It is likelythat the age of the interviewer may initially have been abarrier to the adolescents being open. Nonetheless, itwas not feasible to cover the costs of a younger inter-viewer for subsequent interviews. Relying only on oneinterviewer renders age and gender matching impossiblein dyadic research of this nature.We took several measures to minimise the power dif-

ferential between the interviewer and participants.Firstly, conducting repeat interviews with the same par-ticipant promoted deeper reflection and trust over thecourse of the data collection. Secondly, interviews tookplace in the participant’s home but after a period inwhich the interviewer would chat informally with theparticipant and observe their daily activities. This en-abled greater access to the contextual details of theirdaily lives, including communication styles and behav-iour patterns. The interviewer was thus better placed tointegrate specific and detailed questions about the rela-tional context of the participant’s lives. This investmentin understanding the experiences of participants in theirsocial context further consolidated trust and rapport.Lastly, to ameliorate the power differential between theinterviewer and the adolescent participants, special at-tention was paid to avoiding controlling behaviours thatmight have led them to associate the interviewer withwhat might typically be expected of an interaction with amother or aunt. Where possible the interviewer gaveconsiderable autonomy to the participants, includinginitiating the scheduling of subsequent interviews sothat they were conducted at a time that suited theadolescents and placing considerable emphasis on thecontrol the participants could exert over whether and

Gumede et al. BMC Medical Ethics (2019) 20:47 Page 8 of 12

how they wanted to respond to any of the questions. Allof this contributed to an overall investment in demon-strating how confidentiality was being maintained whichfurther improved the trust between the adolescents andinterviewer.

Unintended preferential treatment of participants by theresearcherWhile the interviewer spent a significant amount of timein the field to study the lives of the dyads within theirnaturalistic setting, unintended preferential bondingwith individuals, particularly older carers, occurred. Theinterviewer established stronger rapport and more con-nections with the older carers, compared to the adoles-cents. Approaching older carers first and spending moretime at the home waiting for the adolescents contributedto preferential bonding with the older carers. At everyvisit to the home, the interviewer had to announce herpresence to the older carers first and to request permis-sion to speak to the adolescents. Even when the adoles-cents knew the interviewer had come to see them, theywaited to be called by their older carers that the inter-viewer had arrived for the interview with the adoles-cents. In addition, at the end of interviews with theadolescents, the interviewer had to announce her depart-ure to the older carers. This is a social protocol in thearea for home entry and home exit, and it is regarded asa symbol of respect to the elders or heads of families. Assuch, opportunities for interaction between the inter-viewer and older carers were greater than with the ado-lescents. During those interactions, the older carersalways had important and interesting stories to tell theinterviewer, also creating the need to follow-up their stor-ies in subsequent visits. While the interviewer spent ex-tended time with older carers and this generated detailedaccounts about their worlds; however, it compromised ad-olescents’ voices as older carers’ views dominated the ado-lescents’ in the research process. Apart from complyingwith social protocols, our flexible interview structureallowed the interviewer to do the interview with any indi-vidual available between adolescents and the older carers.The interviewer also employed reflexivity as a tool toguide ethical practice throughout the study.

DiscussionIn this paper, we have shared our experience of conduct-ing qualitative dyadic research with adolescent-oldercarer dyads in rural South Africa and presented ethicaland methodological challenges, which have arisen, fromour work. These include challenges on recruitment ofdyads, consenting dyads, confidentiality, conducting sep-arate dyadic interviews, and interviewer-dyad inter-action. We also illustrated how we dealt with thesechallenges.

As guided by the relational agency, recruitment strat-egies are influenced by relationships between dyad mem-bers and the socio-ecological context of the researchsetting. Researchers must consider the cultural values ofthe participants they intend to recruit [34]. We recruitedolder carers first to align with socially accepted practicesto engage with the caregivers prior to the young people.It was considered disrespectful to approach adolescentsabout participating in a research study without havingfirst obtained permission from their caregivers. However,as other authors pointed out, this strategy may not ad-dress ethical concerns about the risk of coercion to par-ticipate among vulnerable populations [4, 5]. Differentapproaches could be employed to deal with ethical di-lemmas arising from the recruitment of dyads. Youngpeople can be approached first [5] or together with theircaregivers [4]. In Uganda, young people nominated theircaregivers and provided permission for researchers toapproach caregivers [4]. By requesting children to iden-tify their caregivers, it eliminates controlling behavioursfrom researchers and facilitates children empowermentin research by giving them a choice and a voice in theselection of dyad members.Confidentiality is one of the cornerstones of research

involving human participants. Protecting research partic-ipants’ right to confidentiality is a responsibility sharedby researchers, institutional review boards, and partici-pants themselves. Interviewing dyads within one intim-ate relationship posed three confidentiality challenges inthis study: limited trust in the confidentiality process,unintended disclosure of information by the interviewer,and right to access confidential information of the otherdyad member.One strength of separate interviews is that it allows

participants more freedom to express their individualperspective than in the absence of their partner [26]. In-terviews with individual partners enabled them to revealinformation to the interviewer while keeping it secretfrom their partner (e.g. pregnancy, sexual relationships).This enhanced our contextual understanding of theinterpersonal relationships of dyads who kept secretsfrom each other [1]. Yet, some participants had limitedtrust in the confidentiality process, regardless of beinginterviewed separately. They anticipated that their infor-mation could be shared with their partners hence theyinsisted on assurance for confidentiality. Our finding re-flects those of Allan who also found that individuals willreveal confidential information if they are certain it willnever be disclosed to their partner [2]. In addition, the re-lational agency states that individuals’ experiences areshaped and influenced by their relationships with others[33]. An individual can have a sense of intentionally or un-intentionally influencing another person in a constructiveor deconstructive way [35]. In our study, the participants

Gumede et al. BMC Medical Ethics (2019) 20:47 Page 9 of 12

had a sense that if the information was disclosed to theirpartners, this could influence their relationships in a de-constructive way. While the adolescents may be narratingsomething to the interviewer, they sometimes appeared tobe economising on what they disclosed to the interviewerperhaps because they are wondering about who will haveaccess to their information. Certainly, in the first inter-views, they may have wondered if they were speaking tothe interviewer or speaking to their older carers by proxyas information might slip through the interviewer to theolder carer.Consistent with the literature, we found that, in separ-

ate interviews, participants expressed the desire to knowwhat their partners said [25, 26]. A number of studieshave begun to examine possible strategies for re-searchers to maintain confidentiality when conductingseparate interviews with dyads [14, 25]. Our strategy re-flects that of Zarhin [25] who also emphasized refrainingfrom revealing information to protect participants’confidentiality. In contrast to our strategy of remind-ing participants verbally about the confidentialityclause, Taylor and de Vocht [14] suggest incorporatingthis clause into the participant information sheets tomake it clear to individuals that “no informationshared by individuals would be disclosed to theirpartner.” However, attempting to be given the right toaccess information pertaining to their partners be-cause of authority over the other is a relational issue.One’s actions and perceptions of self as an agent de-pends on the social and relationship context in whichagency is enacted and experienced [35].Critical to the success of dyad research is the con-

fidence that the participants have in the confidential-ity of the process, which relates not only to the datacollection process but also to the presentation offindings. Securing this confidence is a process ratherthan a one-off activity. It can be secured as part ofthe initial informed consent process, but confidencemay also be secured through the practice of confi-dentiality from recruitment to the dissemination offindings.In this study, we found a single-sided account of

stories between dyad members when interviewed sep-arately. For example, older carers disclosed their HIV-positive status, which their adolescent grandchildrenwere discreet about. This finding was also reported byNorlyk and colleagues [27] when they conducted re-peated interviews with patients and their partnerswho were living with Parkinson’s disease. They re-ported that ethical and methodological considerationswere intertwined when one partner address an issueof interest and the other not [27]. Consistent with theliterature, our study was unable to probe the specificreasons for the adolescents to be discreet about their

older carers’ HIV status, as this may have breachedthe confidentiality of older carers’ information. Theboundaries of individuals were respected because itwas their decision about what and what not to sharewith the researcher [14]. Drawing from the relationalagency, adolescents’ actions are influenced by the na-ture of the relationship with their carers, and the ac-tions are a product of that relationship. Asmentioned, the relational agency explains how adoles-cents’ experiences are shaped by their relationshipsand how their behaviours are influenced by the rela-tional agency. Being silent about their older carers’HIV-positive status might also illustrate how caremight be done in these cases, through discretion. Thisshows the importance of listening to both sides byhaving separate dyadic interviews. Dyad members mayexperience the same event differently; thus, each indi-vidual’s story produces a dyad story, which in turnprovides useful data to understand the dyad relationship.The fact of a single-sided account of a story reveals thenature of the relationship between individuals in the dyad.The implication of this tells us a little about howcaring is bidirectional, but also can be hidden or nottalked about for relational reasons. This methodo-logical approach gives us valuable insights that maynot have been revealed using a survey method orrelying on only interviewing one individual withinthe dyad.Therefore, the methodological and ethical challenges

of the separate interviewing approach are associatedwith the relational agency. They are both a feature of theexperience and a feature of the relational agency. Ratherthan to try to design a method, which resolves all thesemethodological and ethical dilemmas, it is important torecognise these challenges as features, or characteristics,of relationships and to illuminate that they exist. Thechallenges presented in this study displayed how thingsare experienced rather than them being the weaknessesof the method. They capture the essence of the experi-ence of conducting a dyad study with adolescents andtheir older carers in our setting.Although providing some very useful data on this

under-researched issue, the major limitations of thisstudy were the selection bias and the sample size. Priorto the study, we had planned to minimize the selectionbias by creating a sampling frame of all adolescent re-cipients who were cared for by older carers using theimplementing organization’s records. However, theimplementing organisation did not have a database ofadolescent recipients in the manner required to pro-duce the sample frame of those in the care of oldercarers. The sample consisted exclusively of womenolder carers, which means that we should be cautiousabout interpreting methodological and ethical

Gumede et al. BMC Medical Ethics (2019) 20:47 Page 10 of 12

challenges in a dyadic approach involving different dyadgroups. Although providing a unique perspective onethical and methodological complexities in dyadic re-search with care dyads of adolescents and their oldercarers, research involving men older carers could de-velop and enhance our scientific knowledge.

ConclusionThis paper demonstrates our experiences of methodo-logical and ethical complexities associated with a quali-tative dyadic approach in rural South Africa. It showsthe difficulties concerning the recruitment of dyads, con-senting dyads, maintaining confidentiality and conduct-ing separate interviews with dyads; and illustrates howwe dealt with the dilemmas. Despite methodological andethical complexities, this approach allowed participantsto share their intimate experiences, permitted adoles-cents’ accounts to be heard unmediated by older carers’participation and has allowed a unique insight into howadolescents navigate their relationships with older carers.Lastly, we agree these ethical and methodological chal-lenges should be recognised as features of relationshipsbetween dyads rather than weaknesses of the method.

AbbreviationsAHRI: Africa Health Research Institute; LMCIs: Low-and-Middle-IncomeCountries

AcknowledgementsWe wish to thank all the participants and the implementing organisationwho contributed with their time and effort to the study. We thank Prof AnnaMeyer-Weitz for her study supervision and support and Dr. Kobus Herbst forassistance with transport at the beginning of data collection.

Authors’ contributionsDG conceived the study and its design in consultation with JS and preparedthe first draft of the manuscript. SB, NN, SN and JS provided comments ondrafts, and read, edited and approved the final manuscript.

FundingThis work is based on the research supported by the National ResearchFoundation of South Africa (Grant Numbers: 113422) and the Wellcome Trust coregrant 082384/Z/07/Z to the Africa Health Research Institute. Opinions, findingsand conclusions or recommendations expressed in this manuscript are that of theauthors, and the funders accept no liability whatsoever in this regard.

Availability of data and materialsThe datasets generated and analysed during the current study are availablefrom the corresponding author on reasonable request.

Ethics approval and consent to participateThe study received ethics approval from the University of KwaZulu-NatalHumanities and Social Sciences Research Ethics Committee (ref HSS/1109/017D).Written informed consent was obtained from all participants before they took partin the study. For adolescents aged less than 18 years, we obtained written consentfrom their older carers and written assent from the young person themselves.

Consent for publicationNot applicable.

Competing interestsThe authors declared no conflicts of interest with respect to the authorshipand/or publication of this article. This paper presents some findings from thefirst author’s PhD study.

Author details1Africa Health Research Institute, Durban, South Africa. 2School of AppliedHuman Sciences, University of KwaZulu-Natal, Durban, South Africa. 3MedicalResearch Council/ Uganda Virus Research Institute and LSHTM UgandaResearch Unit, P.O Box 49, Entebbe, Uganda. 4Sydney School of PublicHealth, The University of Sydney, 324, Edward Ford Building A27, Sydney,Australia. 5Departsment of Global Health and Development, London Schoolof Hygiene and Tropical Medicine, London WC1E 7HT, UK.

Received: 5 December 2018 Accepted: 19 June 2019

References1. Eisikovits Z, Koren C. Approaches to and outcomes of dyadic interview

analysis. Qual Health Res. 2010;20(12):1642–55.2. Allan G. A note on interviewing spouses together. J Marriage Fam. 1980;

42(1):205–10.3. Triado C, Feliciano V, Sole C, Osuna M-J, Pinazo S. The meaning of

grandparenthood: do adolescent grandchildren perceive the relationshipand role in the same way as their grandparents do? J Intergener Relatsh.2005;3(2):101–21.

4. Namukwaya S, Paparini S, Seeley J, Bernays S. “How do we start? And howwill they react?” disclosing to young people with perinatally acquired HIV inUganda. Front Public Heal. 2017;5(343):1–10.

5. Kawuma R, Bernays S, Siu G, Rhodes T, Seeley J. “Children will always bechildren”: exploring perceptions and experiences of HIV-positive childrenwho may not take their treatment and why they may not tell. African JAIDS Res. 2014;13(2):189–95.

6. Phillips EA, Comeau DL, Pisa PT, Stein AD, Norris SA. Perceptions of diet,physical activity, and obesity-related health among black daughter-motherpairs in Soweto, South Africa: a qualitative study. BMC Public Health. 2016;16(750):1–10.

7. Caldwell K. Dyadic interviewing: a technique valuing interdependence ininterviews with individuals with intellectual disabilities. Qual Res. 2014;14(4):488–507.

8. Gysels M, Shipman C, Higginson IJ. Is the qualitative research interview anacceptable medium for research with palliative care patients and carers?BMC Med Ethics. 2008;9(7):1–6.

9. Kendall M, Murray SA, Carduff E, Worth A, Harris F, Lloyd A, et al. Use ofmultiperspective qualitative interviews to understand patients’ and carers’beliefs, experiences, and needs. BMJ. 2009;339(b4122):1–7.

10. Morris SM. Joint and individual interviewing in the context of cancer. QualHealth Res. 2001;11(4):553–67.

11. Hertz R. Separate but simultaneous interviewing of husbands and wives:making sense of their stories. Qual Inq. 1995;1(4):429–51.

12. Voltelen B, Konradsen H, Ostergaard B. Ethical considerations whenconducting joint interviews with close relatives or family: an integrativereview. Scand J Caring Sci. 2017:1–12.

13. Polak L, Green J. Using joint interviews to add analytic value. Qual HealthRes. 2016;26(12):1638–48.

14. Taylor B, de Vocht H. Interviewing separately or as couples? Considerationsof authenticity of method. Qual Health Res. 2011;21(11):1576–87.

15. Bjornholt M, Farstad GR. “Am I rambling?”: on the advantages ofinterviewing couples together. Qual Res. 2012:1–17.

16. Forbat L, Henderson J. “Stuck in the middle with you”: the ethics andprocess of qualitative research with two people in an intimate relationship.Qual Health Res. 2003;13(10):1453–62.

17. Thomas N, O’Kane C. The ethics of participatory research with children.Child Soc. 1998;12:336–48.

18. Knottnerus JA, Tugwell P. Ethics of research methodology requires amethodology of research ethics. J Clin Epidemiol. 2018;100:v–vi.

19. Shabalala F, De Lannoy A, Moyer E, Reis R. Rethinking the family in thecontext of care for adolescents living with HIV in Swaziland. AIDS Care.2016;28(sup4):8–17.

20. Nyambedha EO. Ethical dilemmas of social science research on AIDS andorphanhood in Western Kenya. Soc Sci Med. 2008;67:771–9.

21. Abebe T. Multiple methods, complex dilemmas: negotiating socio-ethicalspaces in participatory research with disadvantaged children. Child Geogr.2009;7(4):451–65.

22. London L, Tangwa G, Matchaba-Hove R, Mkhize N, Nwabueze R, Nyika A,et al. Ethics in occupational health: deliberations of an international

Gumede et al. BMC Medical Ethics (2019) 20:47 Page 11 of 12

workgroup addressing challenges in an African context. BMC Med Ethics.2014;15(48):1–11.

23. Osamor PE, Grady C. Autonomy and couples’ joint decision-making inhealthcare. BMC Med Ethics. 2018;19(3):1–8.

24. Francis JK, Fraiz LD, de Roche AM, Catallozzi M, Breitkopf CR, Rosenthal SL.Management of adolescent-parent dyads’ discordance for willingness toparticipate in a reproductive health clinical trial. J Empir Res Hum Res Ethics.2018;13(1):42–9.

25. Zarhin D. Conducting joint interviews with couples: ethical andmethodological challenges. Qual Health Res. 2018;28(5):844–54.

26. Valentine G. Doing household research: interviewing couples together andapart. Area. 1999;31(1):67–74.

27. Norlyk A, Haahr A, Hall E. Interviewing with or without the partner present? –an underexposed dilemma between ethics and methodology in nursingresearch. J Adv Nurs. 2016;72(4):936–45.

28. Ummel D, Achille M. How not to let secrets out when conductingqualitative research with dyads. Qual Health Res. 2016;26(6):807–15.

29. Rutakumwa R, Zalwango F, Richards E, Seeley J. Exploring the carerelationship between grandparents/older carers and children infected withHIV in South-Western Uganda: implications for care for both the childrenand their older carers. Int J Environ Res Public Health. 2015;12:2120–34.

30. Schatz EJ. “Taking care of my own blood ”: older women’s relationships totheir households in rural south Africa1. Scand J Public Heal Suppl. 2007;69:147–54.

31. Kasedde S, Doyle AM, Seeley JA, Ross DA. They are not always a burden:older people and child fostering in Uganda during the HIV epidemic. SocSci Med. 2014;113:161–8.

32. Kuo C, Operario D. Caring for AIDS-orphaned children: an exploratory studyof challenges faced by carers in KwaZulu-Natal, South Africa. VulnerableChild Youth Stud. 2010;5(4):344–52.

33. Bernays S, Bukenya D, Thompson C, Ssembajja F, Seeley J. Being an‘adolescent’: the consequences of gendered risks for young people in rural.Childhood. 2017:1–15.

34. Wittenborn AK, Dolbin-MacNab ML, Keiley MK. Dyadic research in marriageand family therapy: methodological considerations. J Marital Fam Ther.2013;39(1):5–16.

35. De Mol J, Reijmers E, Verhofstadt L, Kuczynski L. Reconstructing a sense ofrelational Agency in Family Therapy. Aust New Zeal J Fam Ther. 2018;39:54–66.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Gumede et al. BMC Medical Ethics (2019) 20:47 Page 12 of 12