Which primary care model? A qualitative analysis of ward-based outreach teams in South Africa

Abstract Globally, models of extending universal health coverage through primary care are influenced by country-specific systems of health care and disease management. In 2015 a rapid assessment of the ward-based outreach component of primary care reengineering was commissioned to understand implementation and rollout challenges. Aim This article aims to describe middle- and lower-level managers’ understanding of ward-based outreach teams (WBOTs) and the problems of authority, jurisdiction and practical functioning that arise from the way the model is constructed and has been operationalised. Setting Data are drawn from a rapid assessment of National Health Insurance (NHI) pilot sites in seven provinces. Methods The study used a modified version of CASCADE. Peer-review teams of public health researchers and district/sub-district managers collected data in two sites per province between March and July 2015. Results Respondents unequivocally support the strategy to extend primary health care services to people in their homes and communities both because it is responsive to the family context of individual health and because it reaches marginal people. They, however, identify critical issues that arise from basing WBOTs in facilities, including unspecific team leadership, inadequate supervision, poorly constituted teams, limited community reach and serious infrastructural and material under-provision. Conclusion Many of the shortcomings of a facility-based extension model can be addressed by an independently resourced, geographic, community-based model of fully constituted teams that are clinically and organisationally supported in an integrated district health system. However, a community-oriented primary care approach will still have to grapple with overarching framework problems.


Introduction
Community-based primary health care is increasingly recognised as an essential part of a strong health care delivery system. 1 It is seen as a sustainable and affordable way of achieving public health goals across all segments of society. 2 While efforts to achieve universal health coverage through primary care all involve community health workers (CHWs) of some description, actual intervention models around the globe vary quite considerably. 3 Not only are they place and time specific, but they are also strongly influenced by historical and contemporary country-specific practices and their embedded assumptions. This is true for South Africa. National health care system reform was initiated by the South African government in 2011. 4 It includes community-based outreach services, school health services, effective referral systems and improved emergency and planned patient transport services delivered through purposebuilt teams. The intention is that district specialist teams improve facility functioning, and the school health and ward teams extend existing facility services through outreach to schools and communities. 5 In this model these and all other tiers and divisions of health care provision are to be integrated through a National Health Insurance (NHI) financing system. http://www.phcfm.org Open Access understand reform challenges and to identify ways of developing and scaling the approach beyond the NHI pilot sites.
This article presents findings on middle-and lower-level managers' (district and sub-district managers, facility managers and WBOT leaders) understanding of the purpose and approach to WBOTs and the problems of authority, jurisdiction and practical functioning that arise from the way the model is constructed and has been operationalised.

Research methods and design
The rapid assessment used a modified version of CASCADE, 6 a qualitative peer-review methodology developed by the European Union-funded Cities Exchanging on Local Energy Leadership programme. It is designed to learn from and strengthen sites and systems by promoting equity in learning and assessment.
Research for this study was undertaken in two sub-districts in each NHI pilot district in seven provinces according to a prescribed work plan (Box 1). At the request of the National Department of Health (NDoH), three 'best practice' initiatives (City of Tshwane/UP COPC Intervention; Philani Maternal Child Health and Nutrition Project; and Community Action and Palliative Care) were included as case studies in the rapid assessment. They were either not in an NHI pilot site or they were run by the local municipality in an NHI pilot site but not included in the assessment's terms of reference.
Study sites were purposively selected with the assistance of district health staff. The best and the worst performing sub-district in respect of primary care reengineering were included in the research.
Two research teams (one from each of the participating universities) undertook field research, visiting each site in their designated provinces. Each team comprised two to five university researchers and two or more health care managers, professionals and/or workers selected from study sites. Research teams spent five days in each assigned sub-district.
Data were collected using a purposively designed set of 'benchmarks for integrated learning' that guided all fieldwork conducted by the research teams. The benchmarks focused on five key topics -mindset/approach, preparation, implementation, networking/relationships and monitoring/ evaluation. These topics were used to design a pre-visit questionnaire. The benchmarks and sub-questions were also used to guide in field observations, key informant interviews and focus group discussions at each site.
The questionnaire was distributed to district and sub-district managers as well as team leaders prior to fieldwork. It was completed by respondents and collected during site visits. The findings presented here are based on an analysis of questions in the first three benchmark topics that relate to the article's focus -drawing either from responses in the questionnaire (for the NHI sites) or in respect of the case study, from document reviews and key informant interviews. A total of 44 questionnaires were returned and four key informant interviews conducted.
Questionnaire responses were captured into MS Excel. The benchmarks and questions provided a clear framework for analysis. The authors intensively read and discussed the data, leading to an identification of themes, common issues and exceptions. Quotations best illustrative of the evidence for the arguments are provided in the next section.

Ethical consideration
Ethical approval to conduct this study was obtained from the research ethics committees of the University of Fort Hare (REC-270710-028-RALevel01) and the University of Pretoria (Ethics Reference Number: 102/2011 Amendment 26062014).

Approach and purpose
In terms of the approach and purpose of ward-based health care, respondents across most study districts understand the national model as one of clinic-based, primary health care service extension. Respondents distinguish the WBOT approach from the legacy model of facility-based primary health care service, which they describe as follows: 'Primary health care was normally seen as the entry level of service that the community accessed through a primary health care facility. Patients had to come to the facility to get a service. District services from clinic staff to households were extremely limited and services therefore were mainly geared at patients coming to the facility.' ( WBOTs are expected to make it possible to extend health care to people who are on the social or physical margins of the health care system. In Tshwane Inner City, for instance, in addition to going to households, WBOTs go out to people who are homeless, abuse substances, engage in sex work or are forced to the fringe of society in other ways (Dr Mohale M 2015, in-depth interview, August 28).
They also aim to reach rural and remote communities. 'In primary health care people were defaulting treatment because of bad roads and the long distances they used to have to travel. The WBOT team is able to visit households in the community as well as screen children in schools.' (Eastern Cape, SA RES3, Female)

Team composition and system functioning
Respondents say the model is one in which services are delivered by teams made up of: 'One professional nurse, six or more CHWs, one health promoter and one environmental officer.' (Mpumalanga, SA RES2, Female) and that WBOT personnel will be recruited into or deployed from within the health system and government-funded NGOs.
In terms of operationalising WBOTs according to the NDoH model, respondents describe model-related challenges of functional authority and jurisdiction as well as implementation challenges of system preparation and understanding, human resourcing, infrastructure and materials.
Respondents report that WBOTs are poorly understood at various levels of management in the public health care system. In terms of human resourcing all study sites experience common interrelated issues. WBOTs are supposed to be led by professional nurses.
Health system managers most often assign facility-based professional nurses as WBOT team leaders. In part they take this route because recruitment into the system is stymied by a dearth of professional nurses. In addition to team leadership challenges, respondents also report problems with team composition. WBOTs are supposed to be made up of at least eight community health and home-based care workers. Several respondents report that there are either too few or too many CHWs in teams. They say that this affects both the ability of WBOTs to provide services and the ability of team leaders to properly manage their teams.

Infrastructure and resourcing
In all provinces respondents say that attention to infrastructure and material resourcing of ward-based primary health services is inadequate. They say that NHI grants are both restrictive and subject to competing interests. 'There is no clear strategy from a provincial level -no dedicated budget to implement.
[We] use the NHI grant for certain aspects, but it is very prescriptive and cannot be used for anything.' (Northern Cape SA RES5 Female) As a consequence, districts and facilities have had to devise solutions from within already limited means. 'There was a lot of pressure on the district to start the service. The district therefore had to work out its own strategy.' (Northern Cape, SA RES5, Female) They describe how these constraints seriously hamper their ability to work in a multiplicity of ways. All respondents report problems of space. Respondents also point to a lack of uniforms, name tags that identify CHWs to the community, protective clothing, properly equipped kitbags and essential equipment.

The strategy and the model
Among respondents there is unequivocal support for the strategy of taking health care to people in their homes.
'It's the best thing ever to happen to our country. Not only are we addressing the health of our people, but most importantly, we are entering their household to identify the possible causes of [disease] and by assessing the community as a whole.' (Limpopo, SA RES1, Female) 'The difference that I see is the fact that for primary health the client has to come to the facility or mobile clinic to access health care, whereas in WBOTs the client gets health care at his household. This service is an economic relief for them especially for disadvantaged communities that are very far from health care.' (Eastern Cape, SA RES8, Female) However, from their experiences of the challenges of the model and its practical implementation they make five key recommendations that relate to the findings presented in this article. They suggest that ward-based health care should be • part of an integrated health care system.
• located outside of clinics.
• staffed to meet service needs (teams with team leaders, professionally supported). • equipped to be functional and effective.

Discussion
In 2010 the National Department of Health 8 introduced primary care reengineering as part of a multi-pronged system reform. In it WBOTs were initiated as a way of extending the reach of existing facilities at the bottom of the health care hierarchy. As the nomenclature 'outreach' suggests, the model is one of creating WBOTs as an add-on service attached to and managed by primary care clinics. With the exception of Tshwane, this is in fact how the model has been practically implemented in all districts.
The extension of facility services model is not unique to South Africa, 9 nor is it unique to the contemporary period.
It also is neither the only possible nor the most appropriate response to the health gap in contemporary South Africa. The 'unsteady march' to improve the health of ordinary people through community-focused health promotion and disease prevention dates back to the community health centre movement first advocated for (UK and USA) and implemented (Soviet Union, Ceylon/Sri-Lanka and China) in the 1920s 10,11 and subsequently expanded and elaborated into community-oriented primary care (COPC) by the Karks in the 1940s in South Africa. 12 Building on an initial Tshwane District (Gauteng Province) pilot (2011-2013), the City of Tshwane, together with the University of Pretoria, has used COPC as an approach to build a community health platform. In it ward health teams are based in health posts and linked to clinics and other district health facilities as well as private sector practices. 13 There are several insurmountable problems that flow from the facility-based extension model. In the first instance there is the difficulty of its physically specific but structurally tangential location in the health care system. By government's own account primary care reengineering was initiated to address the extreme pressures that ill-health places on health care facilities and national well-being in general. 14 It was also introduced to achieve universal health coverage that the existing facility-based, specialist-focused and disease-driven system is unable to address.
As respondents describe, attaching WBOTs to clinics adds additional management and service responsibilities onto already strained, overstretched, under-resourced and underperforming clinics and CHWs. 15,16,17 The model creates functional tensions within the health care system. Team leaders who work within facilities have to juggle the demands of their clinic based responsibilities with the expectation that they simultaneously work with CHWs in the wards. Inevitably, they are inclined to prioritise their facility work at the expense of community health-based activities. It also creates functional tensions between the health care system and the NGOs that formally employ CHWs. Because CHWs work under the clinics but for NGOs, team leaders (and CHWs) have to manage conflicts of authority and jurisdiction. They also have to manage conflicting organisational cultures because in the NGO sector, CHWs frequently work with little or no supervision, whereas in the public health care system there is both a clear hierarchy of authority and an expectation of supervision and oversight.
The model is unable to address the vexing issue of clinical leadership and oversight that is essential to the quality and impact of all health care services and is integral to the intent of universal health coverage. 18 As described, the difficulty of recruiting professional nurses to lead teams and the expectation that those employed in clinics also work in the communities inevitably short-change ward health services and leave CHWs under-or unsupported. Despite government's best intentions, within the existing human resource development framework it is impossible to circumvent the absolute shortage of skilled professional health workers. Like many countries around the world, South Africa faces a growing crisis in human resources for health. 19 It is this crisis that is fuelling a global drive to fundamentally reform national and international approaches to capacity development. 20 Many of these issues relate to who constitutes the health care workforce as well as the education and training they require to be able to deliver quality services. In addition, there are organisational dimensions of the model of delivery that also need to be reformed. One solution that has been proposed (CDC 2016, personal communication, October 6) is to create functional health catchment areas in each district to drive clinician-led inter-professional care in the community health platform. Defined by population, physical geography and existing facility services, each health catchment area should have a team of locally available health and care professionals led by a clinician (a family physician or general practitioner or clinical associate) and should involve both a professional nurse and a social worker. It can also include other professional practitioners (e.g. psychologists, occupational, physical, audio and speech therapists, environmental health officers, nutritionists), depending on their availability. 21 This solution draws on the experiences of implementing COPC with the City of Tshwane and is in keeping with the establishment of district specialist teams to support the facility platform that is part of primary care reengineering.
The problem of insufficient planning, especially around financing and the consequent inadequate material resourcing of WBOTs is related in part to the overall framework and in part to the model. It could have arisen from an assumption at national and provincial level that dedicated financing was not necessary as additional resources would be met from NHI and other grants. This has not happened. Notwithstanding provincial variations 2 it also may be a consequence of inadequate preparation of and engagement with district and facility health care managers and professionals. Respondents mentioned both these possibilities. Other reasons also may be behind the poor planning and resourcing of WBOTs. As in many other countries 1 it might be because of the unfounded assumption that ward-based health care could be implemented at a low cost and within the existing budgets by rearranging personnel, space, equipment and time at the facility. Faced with system expectations and under-resourcing, in practice, this is precisely what clinic and district managers say they have done. They organise and run WBOTs around facility services and needs. As laudable as their pragmatic response is, however, it is unlikely to yield the expected or potential health and economic returns. Generally, the evidence from South Africa 22 and other middle-and low-income countries is that a community-based health platform involving CHWs requires dedicated upfront and recurrent national funding. 1 The facility-based model is unable to adequately address the equity issues that underpin the drive for universal health coverage. Its design and organisation, including having too few teams, having to work from and report at clinics, having to work un-or under-supervised and under-equipped, having to walk long distances or depend on intermittent transport to go to people in remote areas, limit the extent, frequency and quality of community-based health care services. 23 The model even may have given rise to new inequities because it transfers some of the cost burden of health care access to CHWs and team leaders who, in their commitment to deliver outreach, incur workload, time and out-of-pocket transport and communication expenses.
Respondents make very clear recommendations as the summary analysis of issues and recommendations given in Figure 1 clearly shows.
These fit well with the COPC approach that is being put into practice through a City of Tshwane-University of Pretoria partnership. Implemented under the auspices of the NHI and primary care reengineering, the drive to extend universal coverage through a community-based platform has been assigned dedicated City of Tshwane funding. Located outside of facilities, teams work from local health posts with all individuals and families living and working in geographically defined spaces and places. Service and support needs therefore are defined by geographical location rather than disease. Service practices are generalist and comprehensive. Teams of CHWs are clinically led and professionally supported in their daily work by professional nurses, clinical associates, doctors and family physicians. Health care capability is developed and supported through systematic and ongoing work-integrated learning that is designed to continuously expand health literacy and build health care competency.

Conclusions
The rapid assessment of WBOTs found extensive support for primary care reengineering at all study sites in all provinces. However, as this article has shown, respondents do not believe that a facility-based extension model is able to deliver on the promise of universal health coverage that extends to and from the home through communities, facilities and partner services.
In the South African public health context, the COPC approach to universal coverage demonstrably addresses many of the shortcomings of a facility-based extension model. This said, it continues to grapple with systemic problems of vertical programmes, poor health service integration, insufficient and under-prepared capacity, weak inter-sectoral cooperation and the marginal status of CHWs in the health care workforce.
System ambivalence makes it hard for any model to overcome vexing legacy problems as well as the intrinsic system bias towards expensive facility and specialist health care. This is especially the case in terms of financing a community health platform, securing the place of CHWs in the health care system and developing a learning system to meet the skills needs for the 21st century. In a high-stakes context laying the foundation of national wealth on a community health platform that integrates services across sectors and through systems to and from individuals and families in their homes is a low-risk strategy. Having taken the bold step of system reform through primary care reengineering, there is a need to commit to a framework and a model that will realise a strategy that is widely supported by health care providers on the front line and extensively welcomed by people in communities across the country.