Expanding Community Health Worker decision space: impacts of a Participatory Action Research training intervention in a rural South African district

BACKGROUND While recognised in policy and strategy, in practice, Community Health Workers (CHWs) in South Africa experience many challenges. Since the COVID-19 pandemic, CHW roles have expanded, shifting from communities to clinics. The objective was to assess a community-based training intervention to support functionality and local decision-making of CHWs in rural South Africa, aiding CHWs to undertake new, expanded roles during the COVID-19 pandemic. METHODS: CHWs from three rural villages were recruited and trained in rapid Participatory Action Research (PAR) methods via a series of workshops with community stakeholders. Training was designed to support CHWs to convene community groups, raise and/or respond to health concerns, understand concerns from different perspectives, and facilitate and monitor action in communities, health, and other public services. Narrative data from in-depth interviews with CHWs before and after the intervention were thematically analysed using the decision space framework to examine functionality in devolved decision-making. RESULTS: CHWs reported experiencing multiple, intersecting challenges: lack of �nancial, logistical and health systems support, poor role clarity, precarious employment, low and no pay, unstable organizational capacity, and fragile accountability mechanisms. CHWs had considerable commitment and resilience in the face of COVID-19 in terms of increased workloads, increased risk of infection and death, low job security and poor remuneration. The training intervention addressed some resourcing issues, increased management capacity, gave CHWs greater role clarity, improved community mobilisation skills and forged new community and facility-based relationships and alliances. Through regular spaces and processes for cooperative learning and collective action, the intervention supported CHWs to rework their agency in more empowered ways with communities, clinic staff and health managers, and among peers. The training thus served as an implementation support strategy for primary healthcare (PHC). CONCLUSION: The analysis revealed fundamental issues of recognition of CHWs as a permanent, central feature in PHC. The training intervention was positively impactful in widening decision space for CHWs, supporting functionality and agency for local decision-making. The intervention has been recommended for scale-up by the local health authority. Further support for and analysis of how to sustain expansion of CHW decision space is warranted.


Introduction
Community Health Workers (CHWs) play essential roles in primary health care (PHC) in low and middleincome countries (LMICs) [1].They are integral in linking communities and health systems, and in the achievement of Universal Health Coverage (UHC) [2] [3].In practice, however, CHWs experience many challenges undermining capability and potential.This article reports on a community-based intervention to support functionality and local decision-making of CHWs in rural South Africa.
CHWs have long been a part of the South African health system, albeit with variable recognition and support.CHWs in South Africa were rst trained as malaria assistants in the 1920s [4].In the 1940s, despite growing segregation, social medicine advocates identi ed the need for integrated healthcare.This led to establishment of 'health centres', codi ed in the 1942 National Health Service Commission and the 1945 Gluckman Report [4] [5].Health centres were racially biased, however, only 10% catered to black communities [4] [5].Government support subsequently waned, and many health centres closed over the following years [5] [6].The Alma Ata Declaration of PHC in 1978 fostered renewed recognition of CHW programmes [7].While initially successful, resurgence of CHW programmes in the 1970s collapsed postapartheid with the National Health Plan, which discarded the CHW cadre[8] [9] [10].
HIV/AIDS led to a renewed focus and CHW programmes scaled-up, through Home-Based Carers (HBCs).
CHW activities were grounded in 'task-shifting' to support health systems managing the epidemic and severe health worker shortages [11] [12].HBCs worked through community-based organizations (CBOs) in varied roles: HIV counselling and testing; palliative care; Directly Observed Treatment (DOTs) for TB; and caring for children orphaned and disenfranchised by the epidemic [13] [14].Despite government support, lay workers in South Africa continued to face low remuneration, and poor working conditions [15] [16].
Nevertheless, by 2010, there were approximately 70,000 CHWs in the country employed by over 3,000 CBOs [17].
Recently, the district health system (DHS) has become the focus of decentralising reforms in the country [18]: "empowering local administrative units to control their own healthcare agendas and resources… tailoring service provision to the local population" [19].This shifted services to close-to-household care, focused on health prevention, promotion, and community involvement [20].In 2011,a phased transition to a single-payer National Health Insurance (NHI) system was enacted through commitments to UHC, accompanied by a re-engineering of the two-tier system.A three-pronged approach to PHC was progressed comprising: Ward-Based Primary Healthcare Outreach Teams (WBPHCOTs), School Health Services and District Clinical Specialist Teams (DCSTs) with focus on maternal and child health [20].
WBPHCOTs comprise CHWs (6-10 per team), with nurse team leaders connected to local PHC facilities [21].WBPHCOTs cater for populations of approximately 7,600 [22].A WBPHCOT policy framework has been in existence since 2017 to improve working conditions, recruitment, selection, placement, development, management, standardize workplans, ensure standard application and maintain and improve monitoring and evaluation [14].The Department of Health (DoH) reports WBPHCOT successes including extension of healthcare to homeless people, and those in remote, marginalized populations, and extension of care to people lost to follow-up to increase successful treatment completion [14].
There are, however, low levels of integration of CHWs in the formal PHC system and WBPHCOT implementation experiences many challenges [17] [23].A 12-month training curriculum was developed but funding is yet to be disbursed [23].No clear leadership exists at national level, policy governance remains unclear, and investment has been low [23].Corruption is reported as an organizational norm, and health facilities continue to deteriorate [23].CHWs are poorly remunerated and budget allocations are insu cient [23].There is a disconnection in human resources (HR) management between nurse team leaders, employed and paid by the DoH, and CHWs who are paid (and sometimes not) by CBOs contracted by the DoH.More and proper allocation of resources has been called for to strengthen this stream of PHC re-engineering [17] [23].
Despite challenges, CHWs have played a crucial role in COVID-19 [24].Efforts have been hampered, however, by unavailability of Personal Protective Equipment (PPE), poor planning, lack of training and response, absence of incentives and lack of recognition, and in many instances put CHWs at increased risk of infection and death [25].A recent evidence synthesis revealed that CHW roles in emergency pandemic responses rapidly and signi cantly evolve [25].As such, community awareness, engagement and sensitisation are necessary.CHW role clarity, training, supportive supervision, work satisfaction, health and well-being are also critical [26] [27].
CHWs play critical public health roles.Policy support and recognition is unequivocal, as seen in COVID-19, decentralisation and PHC.CHWs, however, need support to execute these roles and realise potential.This article reports on a community-based training intervention with CHWs in rural South Africa.The objective was to assess a training intervention to support and enable CHW to undertake new and expanded roles during the COVID-19 pandemic and within decentralising PHC reforms.

Study setting
The research was based at the Agincourt Health and Socio-Demographic Surveillance System (HDSS) in Mpumalanga; a rural province of 4.7 million [28].Agincourt HDSS is among Southern Africa's oldest and largest population-based cohorts, covering a population of 120,000 in 31 villages over 450km2.Village populations vary from < 5,000 to > 10,000.Orphaned youth characterize the population: school drop-out is 40%, 16% of the provincial population is illiterate, and district unemployment is 37% [29].There are two community health centres and seven PHC clinics [30].The study was nested within the Verbal Autopsy with Participatory Action Research (VAPAR) programme, in which different partners organise evidence for action in a series of action-re ection cycles [31] [32].The VAPAR approach is rooted in health policy and systems research (HPSR) and constructivist and participatory enquiry paradigms, based on assumptions that practical, experiential knowledge that is co-constructed, self-re ective, and embedded in complex, adaptive social and health systems can support and inform the organisation and delivery of equityoriented and people-centred public services.The current study was nested within the third cycle.

Participant recruitment / data collection
As COVID took hold in early 2020, the process was adapted to support the DHS.A rapid, largely virtual consultation, with community stakeholders, government o cials and multisectoral resilience fora, revealed that CHWs were seen as the rst-line response but required support to engage effectively [33].
The third cycle was redesigned to support CHWs with community mobilization through a PAR training intervention.In November 2020, we reviewed the training framework (Supplementary Material 1) with operational managers (OMs) from clinics serving the three study villages.Invitations were then extended to CHWs through OMs.OMs and researchers selected three CHWs based on interest, skills, and motivation, (n = 9 total).Three community stakeholders from previous cycles agreed to join as 'mentors' [34].
An initial workshop engaged PHC staff, CHWs, mentors, OMs, and clinic committee members from the three clinics.Researchers introduced the training, emphasising power-sharing and local action.Through facilitated discussion, participants nominated priority health issues, listed to capture all perspectives and experiences.Lists were then ranked by participants using adhesive stickers.The second and third workshop were held with CHWs and community mentors only.These oriented to the PAR training: collectively problematising, collecting visual data, and facilitating deliberations.CHWs then recruited nine participants from each village (n = 27 total) as: (1) individuals directly affected by issues under investigation, and (2) individuals whose voices might be excluded.
Workshops progressed in each clinic area.There was an alternating sequence of researcher-led training with CHWs (building collective capacities, sharing experiences, with peer-support, re ection, and learning) with CHW and mentor-led sessions with community stakeholders.PAR tools were used to raise and frame local priority concerns and appraise action.Workshops were held in churches, community centres and other community spaces.Mentors facilitated safe spaces free of blame, promoting democratic involvement, and acting as co-facilitators.A nal workshop convened health system and community stakeholders to synthesise ndings, and codesign subsequent multisectoral engagement with local o cials and non-governmental agencies (Table 1).Researchers conducted baseline semi-structured interviews with CHWs applying the 'decision space' framework, exploring the ability to act at community, facility, and district/sub-district levels.End-line interviews examined learning, acceptability, and impact.Interviews were 30-60 minutes long, conducted in the local language, siTsonga by siTsonga-speaking researchers, and structured by pre-prepared topic guides.Data were anonymised, transcribed, and translated to English, and quality checks performed.Audio recordings were reconciled with transcriptions and a percentage of transcripts were backtranslated.Data were encrypted and stored on institutional servers in MSWord, PowerPoint, and image les.There were no refusals in recruitment of CHWs, community mentors and community stakeholders and no dropouts.

Analysis
Analysis sought to assess the extent to which the training intervention supported functionality of local decision-making of CHWs in the contexts of decentralising PHC reforms, and in the context of COVID-19 [35][36].Interview data were the main data source, supplemented by presentations, registers, minutes, eldnotes, and researcher exchanges.Bossert's framework on decentralization in healthcare enables exploration of the extent to which authorities have power to affect decision-making [35] [36].It is structured by three dimensions.Authority refers to clarity of roles and responsibilities, accountability looks at mechanisms of responsibility within and outside the health system and capacity references resources and magnitude to which they use them.Roman et al [37] expanded on Bossert, with an interactive element of the dimensions as de ning decision space [37].Our analytical framework adopted Roman's extension.Apported with NVivo 1, thematic analysis [38] was performed to draw out inductive and deductive themes.Codes were arranged into themes and sub-themes [36].

Ethical considerations
Written consent was gained from all participants, based on written and verbal information and minimum 72 hours for questions to be asked.Anonymity was ensured as was freedom to exit the process at any time.Participants were provided with refreshments and transport costs for workshops, and reimbursed for time participating, 300ZAR (approx.15GBP) per participant for the 12

Results
The thematic analysis is presented below and in Table 2. Non-funded CHWs, i.e., those currently not employed by the government, expressed frustration at the lack of overall support.These CHWs felt that recognition would increase their capacity and effectiveness, especially in the context of the pandemic, and several expressed frustration and concern about unmet needs in communities.All participants had been working at post for a minimum of ve years and had developed skills in patient care at community level.Some noted that career progression for example in nursing was needed for nancial stability and to re ect their value and contribution.The scenario re ected fundamental issues of recognition of CHWs as a permanent, central feature in the DHS and PHC.The PAR intervention was not aimed at employment for CHWs, however, participants reported being equipped to effectively engage the authorities in constructive dialogue and action on such issues.Nevertheless, requests were made for VAPAR to broker these discussions.
(2) Authority: clarity in roles/responsibilities Since the pandemic, CHWs reported a persistent lack of role clarity, high workloads, and top-down management in clinics, all of which limited their agency.CHWs described hierarchical organizational norms, expressed in belittling, and unfair treatment.CHWs felt disrespected and, in some cases, discriminated against by fellow health workers.This included CHWs who had been trained within WBPHCOTs.Untrained CHWs reported that nurses frequently disregarded them during interactions at clinics.In this sense, the shift to clinics together with poor resourcing were major challenges and sent clear and negative messages regarding value and recognition.
While CHWs were initially worried that 'weaknesses' would be exposed in the workshops, during training, they reported gaining authority to raise and frame challenges in a supportive space, through a framework advancing collective mindsets, action, and co-learning.Participants reported that they had learned how to communicate more effectively with supervisors during the training.Participants moreover saw the PAR training, and training in general, as a route to improve their pro les in the system.The PAR training was also seen as bene cial for analytical skills, community engagement, analysis, and public speaking.Participants reported building new, shared understandings about collaboration to achieve shared goals, including occupational.The peer-led spaces aided in bonding, helping CHWs to recognize one another, as well as raise a collective voice to address challenges faced.Through the peer modality, quality relationships between CHWs were seen as a positive experience and strategic bene t.Despite systems challenges, CHWs found the process a positive and 'fun' activity.Overall, these a triple-bene t: community-acceptance, peer support, and recognition in the system.
All CHWs were knowledgeable on their roles and responsibilities.Both trained and untrained CHWs had experience in home-based services.However, untrained CHWs were less clear on new roles and expectations in clinics.There was also uncertainty over what their roles would be in future.The training intervention supported role clarity locally and in the context of shifting roles, as well as recognition of CHWs, and in terms of providing a stable platform to raise awareness and negotiate with actors in higher levels.CHWs recounted that prior to the pandemic, workloads were relatively manageable.They cited abilities to meet targets and previous training as factors enabling them to manage workloads.However, recent demands had signi cantly increased workloads, with regular abandoning of targets and prior frameworks.Some also described a "double burden" with work modalities both in rural communities and clinics, and workloads regularly going above and beyond reasonable expectations.Despite challenges of increased workload combined with lack of training, employment, and logistical support, CHWs expressed deep resilience and commitment.
(3) Accountability: mechanisms of responsibility within and outside system Accountability was consistently narrated as high.CHWs asserted that despite being in a system rife with challenges in resources and organizational capacity to recognize and support CHWs, these affected neither the services they rendered nor the supervisors and system they reported to.All CHWs felt they had overall good working relationships with communities and felt strongly accountable for residents in their care.There were some accounts of detachment from previous roles owing to COVID-19, however, and a sense of loss of contact with communities.Trust was a further theme.Some CHWs reported good levels of trust with communities due to many years working in the rural areas.Others experienced resistance despite efforts to engage.Some CHWs explained this was due to communities' concerns and perceptions over lack of con dentiality over COVID-19 and other communicable diseases.
The training was reported to provide a platform to develop new alliances between communities and services, further build trust and advance collective recognition of and action towards local concerns, such as stigma around communicable diseases.CHWs assessed the intervention to be important in helping to facilitate, convene and mobilize community members to raise awareness, de-stigmatize COVID-19 and other communicable diseases, as well as other health and community concerns.This was seen as strengthening ties that some CHWs already had with communities, and as offering new opportunities to form alliances with those who had challenges with trust.
All participants recounted knowledge of supervision structures at clinics, corroborating ndings on hierarchical norms.CHWs were clear on reporting patterns to communities and clinics.They also felt supported by superiors to carry out duties.They recounted that although they were supported by superiors, they did not always get feedback concerning performance due to overload in the system.
Through the PAR intervention, participants reported an increased understanding of the need to collaborate to achieve shared goals.The workshops provided bonding opportunities for participants and helped them to recognize each other as key agents in PHC provision.Connecting elements of resources, organizational cultures and norms, and accountability processes, the learning platform and training intervention were reported to provide opportunities for CHWs to rework their agency as their value and contribution were explicitly recognized at higher levels of the system.On completion of the cycle, collective re ection with CHWs, community stakeholders and health systems actors (clinic nurses, Outreach Team Leaders [OTLs], OMs and sub district PHC managers) validated ndings on the PAR training intervention and there was a recommendation from the DHS to scale the training outside the study setting [39] Discussion There is widespread recognition of the contribution of CHWs through provision of preventative, promotive and curative services to the progressive realisation of UHC [40].Despite recognition, in practice, CHWs face systemic barriers and biases, which signi cantly undermines potential.There is a clear need to support CHWs to connect with communities, and in health facilities, as part of a whole-of-society approach [41].In this study, CHWs reported a range of inter-related capacity problems: inadequate training, low and no pay, precarious employment, hazardous working conditions, unmanageable workloads, poor career progression, critical equipment shortages, and unstable organizational capacity.
Training was a major issue.CHWs trained by WBPHCOTs had an 'advantage' in clinic settings, while untrained and unemployed CHWs struggled to adapt and be recognised.In terms of authority, there were reports of embedded biases: discrimination, lack of respect, lack of role clarity and limited opportunities for communication and trust-building.While arguably the best-performing domain, accountability mechanisms were fragile.Together, these issues severely restricted CHWs' decision space.
The literature con rms and contextualises these ndings.There are low levels of integration of CHWs in the formal PHC system, implementation has been slow and uneven, and there is low coverage [17][42].
There is a lack of national leadership and nancial support, poor governance, low employment status and pay, political interference, inadequate supervision, and support, particularly in terms of links to facilities, compounded by roles that are poorly de ned and supported [17][42].As of 2020, Mpumalanga DoH has established 235 out of 560 (42%) WBPHCOTs.Nevertheless, the provincial strategic plan 2020-24 aims to absorb all 6,119 CHWs currently funded by CBOs into government contracts [43].
The study assessed a training intervention to support the functionality of CHWs.Initial outcomes suggest a modest but de ned improvement in decision space for CHWs.The analysis demonstrated positive impacts in terms of capacity and authority: provision of vital training resources, greater role clarity, and in improved public speaking and analytical skills.It also contributed to promoting accountability: building trust and communication with communities, in the health system and quality relationships between CHWs.The intervention contributed to widening decision space and thus served as an implementation support strategy for PHC re-engineering.
The analysis highlighted contextual and systems constraints.Challenges such as a fair, living wage, stable employment, career progression, workloads, and logistical support were harder for the intervention to address.These are critical issues that require urgent attention.Through the PAR process, issues surfaced were referred to higher levels, and by virtue of the alliances and relationships built, the positioning and probability of remedial action was optimised.Scaled application establishing and sustaining platforms for dialogue, learning and action is now being rolled out beyond the study setting, replicating, and re ning the model in the rural province.
While the evidence-base is limited, WHO guidance on optimising CHW programmes sets out the need for core competencies in communication and community engagement and mobilisation [44] This study was, in part, organised to contribute to gaps identi ed by Roman et al on understanding and supporting the extent to which authorities have meaningful power for local decision-making [37].The analysis suggests that synergy across the components is vital in enabling or constraining CHW decision space.In this analysis, accountability mechanisms are present but need to be coupled with improved organisational recognition and, critically, improved resources in a range of areas.Interventions that attempt to impact these core components are relevant for overall health system performance (Fig. 1).
Financial, human, administrative, technical, and organizational resources are required to ensure success, scale-up and sustainability and, ultimately, improved local decision-making in the health sector [37][45] [46].
The analysis also contributes to the literature on decentralization and its role in health systems improvement.As Roman et al point out, there will be little change where decision space is unavailable [37].Decentralization has the potential to improve the South African health system experiencing epidemiological transition, with entrenched health inequalities, and an evolving situation in relation to PHC re-engineering, and COVID-19.Widening CHW decision space in this context has the potential to reap great dividends [47].Decentralization however exists to different degrees and in varying extents across different settings [46][48] [49][50].
The analysis suggests that mutually supportive, bottom-up approaches are likely to support PHC reforms and COVID responses, consolidating and harmonising strengths at different levels of the system.'Post-COVID', further challenges play out related to the continued widening and deepening of health inequalities [51], the global human resources for health crisis [52], and the ascendant double burden of infectious and non-communicable disease in LMICs [28] [53].These a rm the need for new forms of real-time health systems and policy learning combining insights from implementation experience with policy and planning [54][55].

Methodological re ections
The analysis was based on a small sample, albeit representative of the surveillance area, and reasonably representative of the province.Positive response bias among respondents may have also been a risk.This was mitigated by triangulating multiple data sources.Deepening the small evaluative evidence base on participatory learning interventions is a priority in future.Decision space is an approach to understand how decentralised health systems operate.It has not been applied in great depth to CHW roles and functions, however, enabled evaluation of a short-term training intervention.In terms of re exivity and positionality, researchers observed and empathized with female CHWs, in their bid to provide for their families and expectations of resilience.
The analysis supported increased understanding of health workers in lower levels of the system, their di culties and bureaucratic challenges faced with regards to decentralization.Capacity building of actors in the health system is warranted in this regard.Otherwise, the HDSS institutional base was a major factor conferring legitimacy on the learning platform and process owing to the long-established presence of and trust relationships with rural communities and the district health system.As stable public health observatories, HDSSs occupy strategically important positions to broker data-driven decisionmaking between people and services.

Conclusion
CHWs face numerous challenges in the health system in terms of their recognition as a permanent, Abbreviations

Table 1
Summary of PAR tools and intended learning outcomes of PAR intervention CHWsAgree overall goal(s) to address issues identi ed and develop stepwise actions to achieve these.Draws on problem tree and Venn diagram to collectively develop moving towards a desired goal via a series of interconnected steps and events VAPAR Re ect on Workshops 11-13 and codesign/plan taking local action plans to multisectoral fora for analysis and planning.Conduct end line interviews.N.B.: Shaded rows indicate community workshops; clear rows denote peer support workshops.Participants Facilitator/s Objectives/tools N.B.: Shaded rows indicate community workshops; clear rows denote peer support workshops.
-week training module.All identi able data were anonymized.Electronic data was stored in password-protected les on secure servers hosted by Agincourt HDSS and University of Aberdeen.The study was approved by Mpumalanga Health Research Committee (MP_201712_003).Protocols were approved by the Research Ethics Committees of University of the Witwatersrand, Johannesburg (M171050), and University of Aberdeen (CERB/2017/4/1457).

Table 2
The shift to clinics during COVID-19 introduced acute logistical challenges: lack of PPE, unavailable transport, and missing equipment at clinics.CHWs often used their own money and resources to ll gaps.CHWs moreover reported inadequate remuneration.Most CHWs were funded and working at facilities under the DoH mandate.However, some still operating under CBOs were unfunded with instances of nonreceipt of stipends since 2015.Other CHWs engaged by clinics had yet to receive remittances for COVID-19 screening at clinics.For those who received some remuneration, it was seen as insubstantial as a All CHWs articulated resource challenges.Training was a major issue.Due to new demands, CHWs urgently needed, and had been assured, clinical training, including for COVID-19.Promises of training were unful lled, however.Most CHWs had basic training and were certi ed carers, however, few were WBPHCOT-trained.This had pronounced impacts expressed in perceptions of low value to, and limited recognition by, the system.CHWs re ected that the PAR intervention was a welcome opportunity to assert themselves as a recognized cadre, through 'role clari cation'.While some felt the PAR tools were challenging, during the workshops, communication, respect, and collective mindsets were illuminated, and relevance of tools to everyday practice was recognized.Nevertheless, the general sentiment was that more training was required and continued training through the PAR intervention was requested.livingwage and that the issue has largely gone unanswered.Despite no and/or inadequate remuneration, CHWs remain committed but interpreted lack of fair pay as further evidence of lack of value.The PAR training intervention provided stipends, travel costs and refreshments for those attending workshops.The project also supported CHWs to broker dialogue with the DoH and other actors in resolving challenges.Nevertheless, CHWs required improved salary support.
[44]commended modalities of training include those that are: theoretical and practical, with priority on supervised practical experience; face-to-face and e-learning, with priority on face-to-face; training in or near the community; in appropriate languages; positive training environments; and inter-professional approaches.There is recognition of the need for evidence on certi cation or contracting and career ladders, as well as contextualized, realistic HPSR understandings of what works, how, with whom, to what extent and within speci c health systems contexts and circumstances[44].