Introduction

The increased use of information and communication technologies (ICTs) for health, also referred to as digital health, is creating opportunities to support maternal and child health. Digital health has the potential to address health inequalities and community health challenges if undertaken in a contextually relevant manner so as not to exacerbate inequities [1–3]. Existing uses in South Africa and other African countries include reminders about antenatal visits, pregnancy registration, child health information, supporting health behaviours through mobile messages, and technologies specifically adopted in response to the coronavirus disease 2019 (COVID-19) pandemic and its impact on health care services [4–9]. Most notably, South Africa introduced a text messaging service called MomConnect in 2014 for the purposes of registering pregnancies and providing health information to pregnant and postpartum women [10, 11].

Participatory approaches have gained popularity in health research, providing opportunities to incorporate contextual insights and preferences of intended beneficiaries throughout the process of designing studies or health interventions [12]. The rationale for doing so includes ethical arguments for democratic involvement of intended beneficiaries, minimising research waste by ensuring the practical relevance of research outputs to stakeholders expected to implement findings, and improving the quality of said outputs through grounding the research process in lived experiences [12, 13]. In South Africa, participatory design and research approaches have been used to, for example, develop interventions for optimising immunisation coverage and service delivery [14], and to support mothers of pre-term infants [15].

A multi-site study, Co-Designing Community-based ICTs Interventions for Maternal and Child Health in South Africa (CoMaCH), was initiated partly due to the unequal nature of digital health development, taking a participatory and community-centred approach to this domain. The project team includes a multidisciplinary, cross-cultural, and cross-geographical consortium of researchers, technology designers, healthcare professionals and community stakeholders who are exploring the potential of ICTs to enhance maternal and child health and wellbeing in South Africa [16, 17].

This article describes the formative qualitative research phase that was undertaken to contextualise and inform the subsequent co-design process [18] in one of the research sites, the urban township setting of Soweto in Johannesburg Metropolitan Municipality. Previous qualitative research on digital health has been limited in settings like Soweto [19, 20]. A better understanding of the socio-technical challenges and opportunities that digital health technologies afford is needed in the context of maternal and child health in Soweto, especially with regards to the COVID-19 pandemic and resulting shifts to digital technologies [9, 21]. The aim was thus to develop a situation analysis of existing services and the practices and preferences of intended end-users, and explore barriers and facilitators to utilising ICTs for community-based solutions to maternal and child health from multiple perspectives in Soweto. An additional aim of the analysis was to generate transferable insights about the ethics and power dynamics of incorporating participant views into planning and design processes.

Materials and methods

The qualitative formative phase involved consulting intended users regarding the context, preferences and previous experiences of digital health. Three groups in Soweto were identified as relevant and feasible to reach for gaining multiple perspectives of the context: recent mothers taking part in a separate health intervention trial run by the research team’s research unit [22]; adjacent caregivers, meaning family members or other individuals with caregiving roles in the mothers’ households; and community health workers employed in the aforementioned trial [23]. Sampling involved purposive considerations of balancing participant groups, and practical considerations of participants’ availability and ability to provide relevant insights. The sample size was pragmatically estimated as a number likely to yield sufficient data for fulfilling the study’s aims. Recruitment was carried out by TM, who invited participants and scheduled interviews. Most participants already had experience of research activities run by the research unit, either through previous studies or having worked as affiliated community health workers. Ethical approval was obtained from the University of the Witwatersrand (M200872, MED20-08-043) and Cardiff University, and written informed consent for participation and audio-recording was given by all participants. Each participant received ZAR 156 (approximately US$ 10) to ensure that no costs (e.g. transport) were incurred from participation.

Semi-structured interviews were conducted at the research unit using interview guides (S1 and S2 Files) shared with CoMaCH research sites in other South African provinces, but with openness to emergent, locally relevant topics. Semi-structured interviews were selected to elicit rich accounts of specific topics of interest, while maintaining flexibility beyond pre-determined questions. As the research unit is situated on a hospital campus, and has the capacity to carry out research following COVID-19 protocols (e.g. symptom screening), these interviews were conducted in-person in line with guidance and approvals from the University. All research staff and participants wore masks at all times, in accordance with South African COVID-19 regulations at the time. The interviews were carried out in private rooms with open windows and sufficient space for physical distancing. MM, GM and SK conducted interviews over two weeks in May 2021 with discussions and debriefs between interviews. Due to MM’s and GM’s working relationships with the community health worker participants, those interviews were carried out in English by SK where possible (7 out of 8 interviews), while MM and GM conducted interviews with ten recent mothers and ten adjacent caregivers, respectively, in English, isiZulu and Setswana as relevant. All interviews were audio-recorded, transcribed verbatim and translated into English where relevant by professional transcribers, and checked and corrected by SK, MM and GM. Audio recordings were stored on a password-protected University server. Copies were securely sent for transcription without any participant identifiers. Some transcription and translation corrections were required, and interpretations arising from the analysis were continuously discussed among the team and checked against recordings to ensure these were not based on incorrectly transcribed or translated data.

To accommodate the multidisciplinary nature of the study, and the three distinct participant groups, qualitative analysis was done using the procedures outlined in the Framework Method [24] approach to thematic analysis [25], which accommodates teamwork and comparisons of different participant sub-groups. The analysis comprised seven stages undertaken collaboratively by the research team: transcription, familiarisation with the interviews, coding, developing a working analytical framework, applying the analytical framework, charting data into the framework matrix, and interpreting the data [24]. SK led the analysis using analysis software MAXQDA, with contributions from MM, GM and TM to coding and the analytical framework, and discussions among the author team regarding the framework matrix and interpreting the data. Coding was initiated through a priori codes based on the interview guide and specific topics that the team had identified as relevant during the interviews. As coding progressed, the initial framework was expanded and refined inductively with new and rephrased codes. Based on the final code framework, themes were developed to capture salient insights and patterns of meaning corresponding with the study aim, and illustrative quotes were charted into the framework matrix to convey nuances and contrasts between different participant groups. The analysis predominantly drew on manifest data content. Higher level abstractions relying on more latent subtleties were treated with caution due to the multilingual and cross-cultural dimensions of the study and discussed in depth among the authors.

This qualitative study employs a subjectivist inductive approach, meaning that the general direction is from context-specific qualitative data to theory and practice [26]. Nevertheless, existing theory and pre-determined topics of interest are also relevant to utilise deductively, and the analysis is thus best described as a pragmatic and theory-informing data analysis, as opposed to being either fully inductive theory development or fully theory-informed [26]. The qualitative approaches utilised here involve assumptions about subjective, local and specific reality and knowledge, while also considering material dimensions (e.g. infrastructure). The study therefore draws on critical realist ontology and epistemology without suggesting a hierarchy of the ontological layers of real, actual and empirical [27, 28].

Procedures to ensure the quality and rigor of the research were undertaken in line with Tracy’s criteria [29] and reported according to the Standards for Reporting Qualitative Research (SRQR) [30], as detailed in the accompanying checklist. Meaningful coherence with the qualitative research paradigm was prioritised over rigid adherence with standardised criteria. For example, we did not pursue inter-rater reliability or member checking, which involve more positivist assumptions about reality and knowledge [31]. Instead, we focused on ensuring the credibility of the analysis through ‘critical friend’ approaches within the research team [31].

Due to the study’s focus on eliciting the views of intended beneficiaries, a relevant interpretive tool is the typology of participation developed by Arnstein through a ladder analogy [32]. The ladder of citizen participation describes different levels from non-participation through degrees of tokenism to the highest degrees of citizen power and shared control over a process. Arnstein’s power-centred theorising has been extended and critiqued in health-related applications [33], but the essence of the ladder as an illustration of power imbalances was taken as a starting point for the analysis of community-based priorities, challenges and facilitators for maternal and child health in the context of societal and health inequalities.

Results

Twenty-eight adult participants were interviewed. There were no refusals to participate, but some interviews were rescheduled due to participants’ other commitments. Ten participants were recent mothers from different neighbourhoods in Soweto, ten were adjacent caregivers (e.g. grandmothers, aunts and sisters from the same household), and eight were community health workers. All participants were Black African women, which reflects the demographics of Soweto and the legacy of colonial and apartheid urban planning. The absence of men as participants is due to both the design of the trial recent mothers and community health workers were recruited from, and the available and actual choices of recent mothers in identifying adjacent caregivers. Through the Framework Method, four themes were developed with different nuances in each participant group, as demonstrated with quotes in Table 1, and pooled descriptions of each theme below.

Download:

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

Table 1. Framework matrix of themes by participant group.

https://doi.org/10.1371/journal.pgph.0001280.t001

The themes are formulated as specific conclusions from the analysis, comprising: 1) coping as a parent is a priority; 2) existing services and initiatives lack consistency, coverage and effective communication; 3) the promise of technology is limited by cost, accessibility and crime; and 4) information is key but difficult to navigate. In addition to these four themes, the dataset captured participants’ ideas and solutions to maternal and child health, and these are described under the sub-heading of ‘Suggested solutions for maternal and child health in Soweto’.

Discussion

This article provides a situation analysis and explores barriers and facilitators to utilising ICTs for community-based solutions to maternal and child health from the perspectives of recent mothers, adjacent caregivers and community health workers in Soweto, South Africa. The qualitative analysis resulted in four themes: 1) coping as a parent is a priority; 2) existing services and initiatives lack consistency, coverage and effective communication; 3) the promise of technology is limited by cost, accessibility and crime; and 4) information is key but difficult to navigate. The message is clear that internet-based technology has great potential for health promotion, but it is not consistently accessed by everyone, and thus may fall short of the need to address rather than exacerbate inequalities [1]. This is in line with amplification theory characterisations of technology as enforcing and amplifying, rather than challenging or transforming, existing institutional or structural forces [34].

The findings also echo previous qualitative research from Soweto in that health concerns, especially any future or relatively abstract health issues, must be seen against the background of more urgent personal needs, such as welfare and income, and underlying structural factors that health interventions rarely address [22, 35, 36]. Coping as parents was a common concern with multiple dimensions such as financial, emotional and practical coping. Participants therefore called for material and social support as well as better health information and improved primary health care services.

The experiences captured in Soweto reflect some broader shortcomings of existing digital health resources, and the challenges of navigating what some authors refer to as information ecology [37, 38]. These findings complement evaluations of MomConnect [39, 40] suggesting that there is a general interest in utilising the service if the issues users are facing can be addressed. Identified technical issues include problems with registering for or consistently using MomConnect, and challenges related to language or literacy when engaging with written information [41, 42]. Our participants expressed more critical views of MomConnect than specific evaluations of the programme seem to have received, suggesting that the participants did not perceive the research team to be representing MomConnect, and felt more comfortable being critical about its shortcomings.

These findings echo recognised impediments to trust in, and utilisation of, digital health technologies, including cost, unequal access, defective technology, poor information quality and inadequate publicity hindering optimal use of already existing digital health solutions [9, 43]. Apart from new solutions, enhancing the quality and reliability of existing services would therefore be a promising and economical avenue for improving digital health promotion in South Africa. It may also be necessary to tailor features of existing services from a user experience perspective, as despite the many requests for a service that can provide health information, participants did not seem to be familiar with such features of MomConnect. Indeed, a national evaluation of the interactive features of South Africa’s MomConnect mobile messaging programme found that only about 8% of registered users engaged with the programme’s features that enabled them to request specific health information [44]. The present study cannot fully explain why utilisation of seemingly promising and acceptable features of existing services remains low, but we recommend engaging users in the development and improvement processes.

While this formative phase involved an openness to emerging ideas and solutions, the scope of solutions was already delineated to digital health, and the topic was broadly described as ‘maternal and child health’, in line with an international funding call. A fully open-ended participatory process for setting the research agenda or intervention focus [45] was not pursued, in part because of pandemic-related delays and challenges. This raises questions about the extent to which intended beneficiaries’ views are listened to and incorporated into the design of research or health interventions [46, 47]. If end-users do not have agenda-setting powers, the scope of solutions is going to be limited to what researchers were able to imagine a priori [32, 48]. This is particularly important to consider in a context where the relatively low priority of health issues vis-à-vis socioeconomic concerns is already well known, and where challenges of technology, including power supply and internet access, easily justify questions like the one posed by a recent mother: “Must you make an app?” (RM5).

Another clash of practical and ethical considerations is the mismatch between participant recommendations, established evidence about health interventions, and problematic histories of paternalistic interventions. For example, participants called for ‘educating’ the community, and addressing health issues through providing information and raising awareness about specific topics. However, it is a typical assumption that giving people information will lead to health-related behaviour change [49, 50]. Based on decades of health psychology, intervention research and implementation science, some recommendations from participants would likely not lead to effective health interventions or solutions. Nevertheless, public awareness is an important component of wider health promotion efforts [49], and participants’ concerns about misinformation or harmful advice should be taken seriously. Finding the balance between listening to participant views, and appropriately incorporating those with existing theoretical and empirical research knowledge is an inevitable challenge of participatory research and health interventions [36, 51, 52]. This, again, points to the responsibility of researchers to also welcome participant views in decision-making, and provide accessible feedback about the process (including why something may not work) in order to be accountable to the groups contributing their time, experiences and ideas to the process.

To return to Arnstein’s ladder of citizen participation [32], unless participatory research or intervention design processes are situated on the highest rungs, with participating communities enjoying full authority in terms of setting the agenda according to their needs and preferences, the decision-making will inevitably be dominated by the more powerful. Intentional efforts by research teams, who tend to hold more power than participants, are needed to start bridging such power imbalances. Researchers engaging in participatory or co-design processes thus have a responsibility to communicate clearly and transparently about the topic, the scope of feasible solutions or the process itself, and the extent to which people’s real concerns can be accommodated within the parameters of the study or intervention design process [48, 53].

In an effort to address problematic power imbalances and avoid paternalistic health interventions, researchers have drawn on Freire’s philosophy [54, 55], aiming to foster a level of critical consciousness about oppressive social structures as a precondition for successful health interventions [56, 57]. Indeed, interventions drawing on notions of critical consciousness have had some success in starting to address structural factors, such as norms around health behaviours, through participatory approaches [58]. Critical consciousness thus enables a recognition of not merely the challenges individuals or communities face, but a deeper understanding of relational factors and power dynamics at play in producing and maintaining (health) inequalities [59]. For developing digital health interventions, critical consciousness could entail awareness and activism around the wider societal inequalities that inevitably influence health promotion efforts, as well as exposure to the more technical and design-specific aspects of digital health [46] (e.g. scale, sustainability and impact [60]). As Fig 1 illustrates, the ladder of participation can incorporate these elements as well, indicating the desired direction towards participation on equal terms. This involves intentional power sharing on the part of the researchers, and critical consciousness on the part of communities, resulting in the ladder analogy becoming more like a staircase.

Download:

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

Fig 1. Staircase of participation [32, 54].

https://doi.org/10.1371/journal.pgph.0001280.g001

The participants provided many potential solutions, both digital and beyond, demonstrating the utility of relatively open-ended qualitative research in informing intervention design. Another strength of the study was drawing on diverse perspectives, both through the different participant groups and the multidisciplinary research team composition, as it enabled a holistic consideration of what may be relevant, effective or feasible. To honour the participants’ contributions, the insights about non-digital solutions are feeding into another Soweto-based research initiative to develop community engagement with health research and interventions [61]. For example, participants’ interest in events where reliable health information is provided can be responded to through the research unit’s wider community engagement and research dissemination efforts, and community gardening initiatives are being piloted and supported by the unit.

Limitations of the study include data quality concerns due to transcription and translation errors, which are challenging to fully mitigate when several languages are used simultaneously. These challenges have been addressed through continuous discussions about reflexivity and interpretations within the author team [62]. Furthermore, the analysis could have benefitted from more contextual and sociodemographic data. Such data could have provided further nuance but were not collected as the study was not designed for sociodemographic representativeness or sub-group analyses. Another shortcoming is the absence of men among participants, limiting the scope of the study in terms of insights about digital health preferences and utilisation of existing services, as these may involve gendered differences [63]. While men’s perspectives were not intentionally excluded, our approach did not involve an active effort to include men, which has been found to be necessary for other qualitative child health research in Soweto [64] and is currently being pursued in other studies at the research unit. The participation and inclusion of men, as well as comparisons between urban and rural research sites, has also been discussed more thoroughly in later phases of the CoMaCH research and co-design process [18].

Conclusions

This study captured insights about the potential of digital and non-digital solutions for maternal and child health in Soweto. Participatory approaches can help realise such potential in more equitable ways, and this study also sheds light on many context-specific barriers to equitable implementation of digital health solutions. These include factors like the cost of data and limited access to devices, such as smartphones, and underlying socioeconomic circumstances that necessitate more holistic consideration and interventions, such as access to welfare and social support. The findings also highlight the value of conducting formative qualitative research to inform co-design processes for developing health interventions. Further research on men’s engagement with digital health in this setting is needed.

Supporting information

Acknowledgments

We would like to thank all the participants in the Soweto study and the CoMaCH UK-South Africa Network members: local and international partners and our cross-disciplinary co-investigators. We gratefully acknowledge the feedback from CoMaCH colleagues to the development of this manuscript.