Methods

Focus group methodology was selected for several reasons. First, focus groups are ideal for understanding the norms and values of culturally diverse populations [30,31]. In India, focus groups have been used to explore a range of HIV-related issues, including factors that may impact on participation in future HIV vaccine trials [32], on acceptability of a vaginal gel among HIV-negative women [33], and on domestic violence on women’s HIV risk [34]. In addition, given the dearth of research on family-based interventions to prevent adolescent HIV infection, focus groups were identified as an ideal methodology to explore the topic with families.

Community background
The study was conducted in Aghai, a village in the Thane district of Maharashtra. Thane, which is north-east of Mumbai and adjacent to Pune, has a population of 8.1 million, of which 30% is rural. In 1986, the School of Social Work at the Tata Institute of Social Sciences established an Integrated Rural Health and Development Project (IRHDP) in Aghai and its 20 surrounding padas, or hamlets. The objectives of the IRHDP are to promote health and education and to effectively utilize and generate local resources for villagers in collaboration with the local primary health centre.

The IRHDP has developed strong community relationships with the local padas. As part of its work, the IRHDP also creates a map of each village and keeps records on the nature of health work conducted in each village. Using the IRHDP village social map and the most recent community census, we selected a pada with which local health workers had a strong existing relationship, but no special history of HIV/AIDS-related work. In total, there were 41 households in the selected pada. Of the 41 households, 25 included at least one unmarried adolescent aged 14 to18 years.

Recruitment and consent
After the sampling frame was finalized, recruitment was conducted via face-to-face outreach by trained, indigenous recruiters who visited homes with eligible adolescents and invited them and their eligible family members to participate. One target adolescent and one target parent from each family were asked to participate. In cases of two or more eligible adolescents, recruiters invited the youngest to participate.

The target parent and adolescent were asked to join a focus group study that sought to understand family members’ perspectives about participating in a family-based programme to help adolescents avoid HIV. As part of the consenting process, families were given basic information related to HIV. Recruiters explained the purpose of the study, the nature of the focus group process, and the right to refuse with no penalty.

A total of 48 individuals were approached to participate in the study and 46 (96%) consented to participate in the study and completed the focus groups. Adolescents received 100 Indian rupees for participating and each parent received 250 Indian rupees (about US$2 and $5, respectively). Institutional Review Board Approval was obtained from both the Tata Institute of Social Sciences (IEC/IRB No: 03/2009) and Columbia University (IRB-AAAC8244); all research protocols complied with the Helsinki Declaration.

Data collection
Separate groups with mothers, fathers, adolescent females and adolescent males were conducted for several reasons. First, Vissandje’e, Abdool, and Dupe’re’ [35] suggest that smaller groups of six to eight participants are ideal for exploring sensitive topics. In addition, triangulating the perspectives of different groups can enhance topic understanding, while homogeneity of group members’ experiences can reduce power differentials and promote participant comfort [36,37]. Finally, gender and age are especially salient factors in some non-Western cultures, where younger persons are discouraged from differing with older or more influential persons, or where females may tend to defer to males [38]. Given these factors, the number of participants per group was kept to six or less.

The standard protocol is to conduct at least three focus groups with each type of participant [36,39]. However, the relatively small size of the population in the village and the high degree of homogeneity of families within and across padas meant that two groups each with adolescent boys, adolescent girls, mothers and fathers were sufficient to cover the research questions. On average, each group lasted for 1.5 hours.

Focus group venues need to be acceptable, private, convenient, and easily accessible for all participants [35,40]. As the pada lacked a common space, the girls and the mothers groups met in the house of the pada worker, and the boys and fathers groups met in the house of the anganwadi (primary school) teacher. The venues were carefully selected spaces that were well known and respected by community members as this was deemed important to engendering participant trust and comfort in the focus group process by the indigenous research staff. Utmost care was taken to ensure privacy during the focus groups. The presence of onlookers and other distractions were minimized by holding the meetings indoors [41,42], and only the focus group facilitators and consented participants were present at each focus group.

Successful focus group implementation depends heavily on the ability of facilitators to moderate the focus group. In this study, the focus group facilitators consisted of the first and fourth authors, and a team of indigenous data collectors. Although all facilitators were familiar with the cultural and demographic profile of the target population, none resided in the target community. The facilitators led each focus group using a protocol developed by the first three authors, and refined with indigenous project staff and community members.

Facilitators then used a “funnel” approach to frame the development of the questioning route [39,43], which allowed for a wider perspective of individual experiences in the initial stages, followed by specific questioning in subsequent stages to directly answer the research questions. This question route enhanced the consistency of data obtained between groups and assisted in efficient, high-quality data analysis [44].

The questions elicited perspectives about the development and implementation of a family-based community intervention for HIV/AIDS in three core domains: (1) perceptions about and preferred format for planned intervention; (2) preferred methods for implementation; and (3) factors that could potentially foster or inhibit full engagement and participation in the intervention. The same sets of questions were asked in each focus group.

Data analysis
Each focus group was tape recorded on an audio cassette and a written verbatim transcript was produced in Marathi. The transcript was translated into English and checked for accuracy using a forward-backward translation method [45]. In addition, the translators reviewed the transcripts to ensure conceptual as well as linguistic equivalence in the translation process [46]. In order to minimize potential bias in data analysis and interpretation, we followed Krueger and Casey’s [36] guidelines to ensure the analysis process was systematic, sequential, verifiable and continuous.

Four independent coders conducted a content analysis to identify “thematic units”, which were defined as frequently occurring sets of explanatory statements [47]. In addition, data were explored for negative incidents and divergent themes [48,49], which added rigour and validity to the results [50,51]. Interrater reliability among the four coders was determined via a frequency count strategy described by Miles and Huberman [49].

Upon completion of coding, each coder independently calculated the frequency that each category and sub-category occurred within the data. The four coders then compared the correspondence in the data analysis. When disagreement occurred, the disagreement was recorded and settled via discussion between the four coders. The total number of agreements was then divided by the total number of agreements and disagreements [49], leading to an interrater reliability of 91%.