The assessment of the outcomes of the feasibility success was based on the objectives. Four broad classifications (process, resources, management and scientific) (Van Teijlingen et al. 2001) were adapted and utilised to map the outcomes of the pilot study (Table 1).

The pilot study was conducted in the rural settlement of Groot Aub in the Khomas Region which is located 45 kilometres from the capital city Windhoek and is easily accessible for data collection. This site was chosen as it has a similar composition to the ones to be chosen for the larger cohort study with respect to variables such as the types of disability, geographic distribution of participants, geographic terrain and cultural diversity. The Groot Aub CBR Programme was initiated in 2015 and the main activities include identification and screening of persons with disabilities, home visits and referral for services. The CBR Programme consists of 15 CBR volunteers who were trained utilising the WHO CBR Matrix (WHO et al. 2010).

The population of this pilot study included all persons with disabilities and their caregivers who are participating in, or are beneficiaries of, the CBR Programme in Groot Aub. The inclusion of caregivers as participants was based on the scoping review (Shumba & Moodley 2017b) that revealed that caregivers can also be utilised to investigate life experiences of persons with disabilities particularly those with multiple and severe disabilities. Purposive sampling utilising a research assistant was used to identify eight participants. These participants had the following characteristics: 18 years or older, involved with the CBR Programme for at least 1 year and able to read and speak basic English. Persons with intellectual disabilities or mental illness or emotional disorders were excluded. The research assistant chosen was the Senior Community Liason Officer who is the Regional Community Liason Officer of the Khomas Region under the Office of the Presidency, Department of Disability Affairs.

Initially 15 participants were invited to participate in the photovoice process, but only eight met the inclusion criteria. Of these eight participants, six completed the study. One of the six participants opted out from group interviewing and requested self-representation of concerns with the local councillor. The participant needed privacy and confidentiality of self-expression for fear of retribution.

Intervention methods included two phases, namely the implementation of the photovoice method and the administration of the WHOQOL-BREF assessment.

Trustworthiness was ensured using Lincoln and Guba’s strategies for the qualitative approach including credibility, transferability, dependability and confirmability (Lincoln & Guba 1985). Credibility was assured by prolonged engagement in photography assignment until the scope of data is adequately covered and data saturation of experiences of participants on the CBR Programme was obtained; referential adequacy through the use of cameras and field notes, triangulation through engagement in individual interviewing, group interviewing and photography, member checks on participants’ responses and peer debriefing of themes with research assistant and second author were also applied. Purposive sampling and dense descriptions ensured transferability. Photographs, written field notes, methodological notes and reflexivity ensured dependability and confirmability.

Ethics approval was obtained from the Human Sciences Ethics Research Committee of the University of KwaZulu-Natal (Reference No: HSS/0646/015D) and the Ministry of Health and Social Services in Namibia (17/3/3). A full disclosure of the purpose of the pilot study was completed including the definition of a pilot study, the objectives of the larger cohort study, the objectives of the pilot study and the criteria for success of feasibility (Appendix 1). Furthermore, the researcher trained the participants to request consent from subjects to be photographed and to publish the photographs for use in conferences, seminars, journals or books (Appendix 2, 3 and 4). Noting the ethical considerations of the photovoice method, the participants were trained on the key concerns and their expected practice during the photovoice process (Appendix 5).

Of the six participants that were included five were persons with disabilities and one was a caregiver to persons with disability. The characteristics of participants are shown in Table 2.

Table 4 depicts the relevant information drawn from integrating the photovoice responses and the WHOQOL-BREF instrument scores. For example convergence is shown with P1 (Participant 1) who had low scores on the Environmental and Physical Health domain and also expressed challenges with lack of water to wash hands at local clinic as well as poor accessibility to toilets. Further divergence is shown with P2 (Participant 2) who had a high score on Physical Health, yet the participant expressed dissatisfaction that she has not yet received proper training on using a white cane, reading Braille, counting money and basic home duties since she became visually impaired.

The overall outcomes were drawn with reference to the feasibility objectives and criteria of the pilot study. The lessons learnt and proposed modifications for the main study are outlined in Table 5.

Although the scoping review (Shumba & Moodley 2017b) demonstrated that caregivers and siblings of persons with disabilities can be included in determining experiences, this study demonstrated that including caregivers and persons with disabilities can affect the outcomes, especially with respect to experiences elicited and measurement of their respective qualities of life using the WHOQOL-BREF instrument. Caregivers and persons with disabilities have different experiences posing challenges in equating their assessments related to quality of life. Furthermore, caregivers introduced issues that are more important to them than persons with disabilities to whom they provide assistance.

The selection of research assistant is key in ensuring quality purposive sampling, competence with local language and retention of participants. In this pilot study, the research assistant was a Senior Community Liaison Officer who had only 1 year experience working with the Groot Aub CBR Programme. This created challenges with purposive sampling as the research team depended heavily on the local persons with disabilities to identify participants. Thus, some of the key attributes in selecting the research assistant in main study are: at least 3 years of CBR experience, at least 2 years working in CBR Programme of that region, should be able to speak a local language of that area and be well versed with local culture and basic sign language skills.

Furthermore, the pilot study excluded persons with hearing impairment as none from the research team had basic sign language skills. Persons with hearing impairment may be good with the photography component, but it may be difficult to engage them in a critical discussion (Jurkowski 2008) as they require sign language interpretation. Another group that this pilot study excluded included persons with intellectual and severe disabilities as a result of the extensive assessments needed and the uncertainty of ensuring that their interest and reasonable accommodation are genuinely taken into account (Ware 2004).

Participants had challenges getting signatures and consent for taking photographs of human beings as most of the participants and subjects could not read and write as indicated in other studies (Akkerman et al. 2014). Consent forms in most cases are written in higher language required by review boards, which may be difficult for participants to understand (Lennox et al. 2005). The researcher translated and explained the consent forms for subjects to be photographed. Furthermore, the researcher recommended the participants to acquire consent by using an ‘X’ from the subjects to be photographed or their caregivers as an indication of signature and then the researcher or research assistant then made follow-ups of these subjects to confirm they gave consent.

The ‘SHOWeD’ technique suggested by Wang in 1999 was not successful with the pilot study as most participants indicated that it was rigid and provided only limited interpretation and exploration of their experiences. In similar studies, McIntyre (2003) also reported that the ‘SHOWeD’ technique had challenges in limiting the participants’ interpretation of their photographs. McIntyre suggested the use of photographer’s instinct in selecting photographs. These photographs are then interpreted and analysed based on personal questions. The researcher then adopted the approach by McIntyre and utilised personal questions. Furthermore, participants were not requested to document their experiences on paper as most participants were not articulate enough to explain clearly their experiences let alone write them concisely.

The main modification of the original photovoice process that was piloted was replacement of collective group discussion with one-on-one interviews in the photovoice analysis. This study revealed that one-on-one interviews with persons with disabilities provided confidentiality allowing participants to express themselves freely without influence from others and preventing the perspectives of participants from being shaped by others (Jurkowski 2008). Furthermore, it provides a more personal setting that facilitates sharing of deeper thoughts with the result being richer interview data (Newman 2010). Also, individual interviewing allowed the researcher to visit each participant’s home and this alleviated the issue of transportation, as most of the participants had mobility or transportation challenges.

To ensure success of the main study, it was key to note the process time for the photography assignment, individual interviewing and to establish the time needed to fill out the WHOQOL-BREF instrument. Although duration of the photovoice method varies according to the objectives of the project, it is important to note that photovoice is more time consuming than other traditional research as it requires developing relationships and trust with participants (Jurkowski 2008). To this end, the researcher should invest a reasonable amount of time and be patient with participants. This study establishes that cameras should be processed immediately after the photography assignment and interviewing should commence as soon as possible so that participants can recall why they took the photo. The WHOQOL-BREF instrument indicated a considerable amount of time required to complete it. Thus, it is ideal to take a short break in between photovoice interviewing and filling in of WHOQOL-BREF instrument to allow participants to refresh.

Photovoice can be used with any disability type provided reasonable accommodation is provided (Shumba & Moodley 2017b). The anticipated resource problems to cater for reasonable accommodation of selected participants include sourcing sign language interpreters and identifying the sourcing and cost of Braille or large print of WHOQOL-BREF instruments. Additional costs to be noted include type of camera to use (digital or disposable-film camera), contingency cameras in case participant camera is lost or broken and distance and transport cost to reach study site. The researcher learnt that it is advisable to actively engage participants throughout the project to increase retention and to avail airtime for mobile phone use to the group leader of participants for constant communication with researcher and research assistant. Furthermore, to help ensure commitment, participants could be given incentives including t-shirts or hats during their photography assignment and certificates of completion at the end of project.

Final stories, themes and theories in photovoice are determined by saturation of data. However, reaching data saturation in the photovoice process can be attained in different ways. It is suggested that participants should be given an opportunity to comment on all photographs that they had taken as limiting the participants to selecting best photographs can hinder the collection and analysis of a wide range of issues (Newman 2010). On the other hand, participants can be given a specified or maximum number of photographs to take in a given period. This allows participants to be critical or selective on which photographs to take. For this pilot study, participants were given disposable cameras (27 exposures film) thus limiting the number of photographs instead of digital cameras, which allow for taking more photographs. Additionally, participants should be trained on photographic techniques to improve picture quality.

Participants were allowed one round of photography assignment and requested to comment on their photographs that best describe their experiences. This was an effective approach as it allowed participants to be critical on issues they photographed and saved time on interviewing and resources needed to process the photographs. However, the researcher noted that participants could be given more than one round of photography assignment. Giving participants more than one round of photography assignment can afford them the opportunity to reflect on their issues and refine their findings. Furthermore, it can allow for saturation when both the researcher and participant can agree that there are no new issues emerging.

The study included the category ‘photographs not taken’ (i.e. photographs participants thought of taking, but failed to, because of other reasons) (Lassetter, Mandleco & Roper 2007) in the feedback session and this produced rich data. One main reason that some participants failed to take some photographs was lack of transport to take photographs of public places where they receive services. Thus, one of the recommendations for the main study is to provide transport arrangements to sites of photography as per participants’ request.

The original conceptualisation of photovoice process by Wang and Burris (1997) allows for group selection of final themes before organising a community gallery. However, Shumba and Moodley (2017b) argued that participants should be given the flexibility to accept or decline proceeding further with the photovoice process after taking photographs. This pilot study gave participants an opportunity to accept or decline the offer to participate in a final group discussion and photo gallery. One participant declined citing confidentiality issues and fear of being victimised by the authorities and family members. Persons with disabilities often lack self-esteem and confidence in representing themselves in a public platform.

Photovoice potentially generates a lot of data, and thus it is important to decide on storage of data and disposal. Several issues need to be taken into account, including: awareness of data protection legislation, secure storage (encryption, lockable locker), awareness of security standards for online or cloud data collection and storage, publication of data and disposal of data (European Commission 2009). When disseminating research findings, it is critical to observe ethical principles including consent for publication of findings. In this pilot study, names of participants and subjects were not used and their faces on photographs were not shown. Furthermore, participants and subjects were asked for consent to publish the photographs in a community gallery, posters, conferences, journals and books.

To determine trustworthiness and data saturation, two or more photography assignments and individual interviews are ideal. However, despite photovoice enabling participants to mitigate communication problems, some of the participants had limited vocabulary and limited articulations skills to explore their experiences and this raised queries on the correctness or accuracy of interpretations of some photographs. In some instances, there was a discrepancy between what the photograph appears to identify and what the participant explained in the interview.

Another issue that is critical to the photovoice process is sample size. The sample of 6–10 was deemed ideal (Wang & Burris 1997). A scoping study (Shumba & Moodley 2017b) reviewed sample sizes from as small as 1 participant to 82 participants. This indicates that photovoice can be used with a range of sample sizes. However, Wang (1999) suggested a group of 6–10 participants as an ideal size for the photovoice method. Such a sample size is considered large enough to offer in-depth experiences and ideas, yet also allows enough time for each participant to contribute in a meaningful way. In addition, groups of this size are small enough so that members are able to feel safe in sharing experiences and taking part in discussions (Palibroda et al. 2009). The current study used a sample size of eight, but only six participants completed the study.

The photovoice process of this study lasted 2 weeks. Duration of photovoice process is determined by data saturation and more than one photovoice assignment can allow for saturation (Shumba & Moodley 2017b). The main study can focus on not only establishing baseline quality of study but also consider longitudinally measuring change in quality of life. The WHOQOL-BREF Group (WHOQOL-BREF Group 1996) proposed some uses of WHOQOL-BREF instrument including establishing baseline scores in a range of areas, determining changes in quality of life over the course of interventions, research and policymaking. Other uses that can be proposed for the main study are determining changes in quality of life at impact evaluation intervals and providing evidence for policy changes (WHOQOL-BREF Group 1996).

Although the WHOQOL-BREF instrument offers a generic measurement on quality of life, WHO developed an additional module to the WHOQOL-BREF called ‘WHOQOL-Dis’ to measure specific aspects on quality of life of persons with physical and intellectual disabilities. Bredemeier et al. (2014) stated that the WHOQOL-Dis is a measurement option for quality of life and thus recommended joint administration with the WHOQOL-BREF instrument. To this end, the main study can potentially jointly administer the WHOQOL-BREF and WHOQOL-Dis instruments to gain both a generic and an in-depth understanding of the quality of life of persons with disabilities.

This study demonstrated that photovoice is a flexible method that can be used with a variety of disabilities and has the potential of being combined with the WHOQOL-BREF instrument. Furthermore, there is need to provide continuous support during the photovoice process and in completing the WHOQOL-BREF instrument. Photovoice allowed most participants who were not confident to verbally communicate to meaningfully participate in the study. Administering the WHOQOL-BREF instrument at the end of the photovoice process can measure both the quality of life of persons with disabilities and indicate the convergence and divergence in the two data collection methods as shown in Table 4. The study demonstrated a stronger convergence than divergence of the two methods (Table 4). Matching the photovoice responses to the scoring on WHOQOL-BREF instrument consistently identified issues that were important to persons with disabilities. Furthermore, integrating the two methods provided an adequate representation of the concerns and issues of persons with disabilities regarding the CBR Programme. However, divergence of outcomes of photovoice and WHOQOL-BREF instrument indicates that caution should be taken when combining both methods because of the diverse nature of persons with disabilities. The participants were requested to identify issues related to all the five components of the WHO CBR Matrix (health, education, livelihood, social relationship and empowerment). There is a potential to explore in-depth issues when the methods are combined to focus on a specific component, for example, health. When there is a specific activity to focus on, the combination is likely to produce in-depth rich data.