Definitions and framework

Decision space is defined as “the range of effective choice that is allowed by the central authorities to be utilised by local authorities” [3] and represents the degree of decentralisation granted to an individual or organisation. This space can be formally defined by laws and regulations, or informally by lack of enforcement of these formal definitions that allows lower level officials at each level to ‘bend the rules’. The assumption is that with increased decision space, managers can make decisions that are more innovative, efficient and responsive to local conditions and that this will improve the quality of service delivery [4].

The paper adopts a conceptual framework (Fig 1) which sees decision space as an emergent property of the authority, capacity and accountability within an organisation, as well as the context in which it operates. Authority provides the de jure decision space by defining roles and responsibilities which enable managers and staff to take action. However, while it is necessary, it is not sufficient, if other features (capacities and accountability) are missing or misaligned. When authority is unclear or fragmented, as is commonly the case, accountability can be undermined [5].

Download:

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

Fig 1. Decision space model.

https://doi.org/10.1371/journal.pone.0304775.g001

Source: adapted from [5]

Capacity is what enables the organisation to function [6], and includes administrative, technical, organisational, financial and human resources. Some of these capacities are inputs, such as finance and staff and valid health system data, but they also include the capacity to manage inputs well. Resources (inputs) are necessary but not sufficient without management capacity, with which they are interdependent [5]. Without resource capacities to ensure organizational functioning, management capacities could also be redundant. Capacity and authority also need to go in step if resources are to be used well. Limited evidence suggests that where resource capacities fall short of what is required for organizational functioning, managers may resort to informal decision-making strategies to fulfil responsibilities and mitigate bureaucratic constraints [4].

Desirable features of capacities link closely to standards for strong health systems generally. For example, financial capacity would be assessed according to adequacy, regularity, flexibility and predictability [7]. For human resources, control over hiring, payment, performance assessment and management, and motivation would be key. For management, technical skills such as planning and management of staff, medicines and supplies and infrastructure are important, as well as having access to relevant information and the ability to use it effectively. Health management information systems produce large amounts of data, yet data are rarely used in local decision-making [8].

In addition, leadership capacity is critical–the ability to create and share organisational vision and motivate staff to adopt it, for example [9]. According to Gilson et al. (2014), PHC will only become a lived reality within the South African health system when front line staff are able to make sense of policy intentions and incorporate them into everyday routines and practices [10]. This requires a leadership of sensemaking that enables front line staff to exercise collective discretionary power to deal with particular contextual needs at the front line of policy delivery.

Accountability refers to answering for decisions or actions, often with the purpose of reducing abuse and improving system performance. It can refer both to the mechanisms for being held responsible for doing the right things and doing them effectively [5, 6], but also the outcome of these. Commonly, a distinction is drawn between bureaucratic and external accountability [11]–the former indicating vertical systems within an organisation, such as planning, target setting, supervision, monitoring, reporting and audits, while the latter is most commonly enacted in the health system through formal community participation mechanisms such as health facility committees and hospital boards. Accountability affects decision space (positively and negatively) and helps to direct its use [12].

Community accountability [13] is an approach to strengthening public accountability through direct involvement of clients, users or the general public in health service delivery. Different measures to enhance community accountability, linked to peripheral facilities, include clinic committees and health interest groups, public report cards and patients’ rights charters. A recent literature review highlighted the way in which bureaucratic accountability mechanisms often constrain the functioning of external accountability mechanisms. Front line managers and providers may be constrained from responding to patient and population priorities due to organizational cultures characterised by supervision and management systems focused on compliance to centrally defined outputs and targets [14]. This may be driven by the need to control misuse of resources, lack of confidence in managers’ capacities, or both. This bureaucratic accountability can crowd out community accountability [15], creating a ‘compliance culture’ which focuses more on tasks than outcomes.

The role of context is also widely recognised as important in relation to these domains and their impact on decision space. Socio-cultural and political factors influence all these relationships, which together affect roles and responsibilities in the health system, its responsiveness and how resources are used [5].

Setting

In post-apartheid South Africa, there was a constitutional commitment to the right to health and community participation for PHC to overcome the historical disadvantages faced by the majority population [16]. Today, significant pro-poor, equity-oriented reforms include: National Health Insurance [16, 17]; PHC Re-engineering [18], which includes the development of Ward-Based Primary Healthcare Outreach Teams decentralising PHC to community level; and the Ideal Clinic initiative, which provides a national quality framework within PHC Re-engineering. Successive national annual performance plans also address workforce development and planning with initiatives on affirmative student recruitment (prioritising students from disadvantaged backgrounds), financial incentives, foreign recruitment and compulsory post training service as well as commitments to strengthen the public sector health workforce through National Health Insurance and the National Development Plan [19].

While policy development is progressive and inclusive, significant gaps exist between policy and implementation in a system characterized by chronic underinvestment, human resource crises, widespread corruption, poor stewardship and deteriorating infrastructure [20, 21]. Health systems also faces a complex ‘quadruple’ burden of socially patterned mortality comprising chronic infectious diseases (HIV/AIDS and TB), non–communicable conditions, maternal and child mortality, and mortality owing to injury and violence [22–25]. The burden of HIV is high and highly unequal. Prevalence in black populations is 40–50 times that of white populations and in adolescents, risks are eight times higher in females than males [20]. South Africa’s child poverty rate is relatively high and its Gini coefficient at 63 is the highest globally, with the majority black population remaining disadvantaged [25]. While early on in the pandemic, South Africa was internationally recognised for timely and decisive action in response to Covid-19, it remains one of the most unequal countries globally, and this is likely to be both reflected in and exacerbated by the pandemic.

Mpumalanga is one of nine provinces in the country, with a population of 4.7 million (7.9% of the national population [26] and a gross domestic product per capita of $12,585 that is close to the national average but well below provinces such as Western Cape and Gauteng [27]. More than half the population is rural [28], while nationally this is around one third [25]. In 2019, provincial unemployment was 35%, with 51% living in poverty [29]. In 2015, life expectancy for males and females was 50 and 53 years respectively, lower than the national average of 60 and 67 years, and under-5 mortality was 41 deaths per 1,000 live births in 2012, which is comparable nationally [30–32].

Mpumalanga Department of Health (DoH) has responsibility to deliver need-based services through an integrated health system covering three districts and 17 sub-districts [30, 33], which include 279 clinics and 33 hospitals [34]. The Mpumalanga DoH structure includes five strategic directorates: HIV/AIDS, sexually transmitted infections and TB Control; communicable disease control; non-communicable diseases; maternal, child, women and youth health and nutrition (MCWYH&N); and research and epidemiology.

Verbal autopsy (VA) data from MRC/Wits-Agincourt Unit’s health and socio-demographic surveillance system (HDSS) in Mpumalanga shows child mortality to have recently declined to similar levels as in the early 1990s. Although this may not seem like a major achievement, on closer examination it clearly has been so, in terms of reversing the disastrous effects of the HIV epidemic on mortality patterns during a period of emerging democracy following decades of apartheid [35, 36]. There are a number of national policies targeting child health, such as the Child Healthcare Problem Identification Programme, which uses child death audits to identify and address immediate and root causes, and the Integrated School Health Policy advancing integrated, holistic approaches to school health [36]. Further progress towards Sustainable Development Goal targets will require widespread improvements in socio-economic conditions and health systems functionality, with national and local leadership commitment.

Within Mpumalanga, we have focused on Ehlanzeni district and Bushbuckridge sub-district, as these are the sites within which the HDSS is located. Ehlanzeni district, the largest of three districts in Mpumalanga with an estimated population of 1.8 million [37]. The district is situated in the south eastern part of the province, bordering Mozambique and Swaziland and consists of four local (sub-district) municipalities.

Design

This paper adopts a qualitative synthesis design [38],drawing on analysis of mixed data sources from 2015 to 2020, which are combined to investigate the domains of our conceptual framework. The first strand draws from the VAPAR programme, which is a partnership of the MRC/Wits-Agincourt Research Unit, Mpumalanga DoH, and collaborating researchers dating back to 2015, to connect community-generated evidence to practitioners, planners and managers through a coproduction, action research process. The second strand draws from experiences of VAPAR team members, and the third from reports, data and documents on health service delivery in the province.

Data sources

The VAPAR process combines VA, a method to determine levels and causes of death in settings where deaths go unrecorded, and participatory action research (PAR), a process in which different stakeholders organise evidence for action. A pilot followed by a series of cycles of PAR was progressed, comprising observation, analysis, planning and action stages.

Community stakeholders were engaged in the initial observation stage from three rural villages, selected based on demographic variation and feasibility, with total populations of between 4,000 and 6,500, within the MRC/Wits-Agincourt Unit’s HDSS study area [39]. Relationships with community stakeholder groups were developed through a series of community-based workshops in which we coproduced new knowledge on under-5 deaths, alcohol and drugs, and water using methods such as ranking, diagramming, stakeholder mapping and participatory photography [40]. A total of 48 community stakeholders participated from three purposively selected villages [41]. In addition, data from the Agincourt HDSS, including VA data on causes and circumstances of death, were integrated into the process [42].

To analyse VA and PAR data and plan feasible action, we held a series of further workshops engaging broadly with government departments, non-governmental agencies and community stakeholders. We jointly identified actions, timescales and implementing partners, culminating in a joint Local Action Plan that all participants committed to. We collected data in presentations, registers, minutes, observational notes and reflective journal data to develop accounts of the process, as well as substantive interpretations and proposed responses.

In the final phase of the cycle, actions were monitored, with cooperative reflection and learning feeding into the following cycle. Those who had committed to specific action items were visited by the researchers at venues of their choice to discuss progress. A structured tool captured mechanisms of change. The process culminated with a collective reflection. Using rapid, participatory methods, we conducted interviews with participants from local communities, government departments and parastatals, non-governmental organisations, and held two workshops with health systems actors. One workshop was held at provincial level with programme and directorate managers, and one at national level, with health programme and policy specialists. We sought perspectives on whether and how impacts had been achieved; acceptability and utility of the process; levels and mechanisms for integration into the health system; and future linkages [41].

In addition to VAPAR data, this paper draws from team discussions which reflected on and elicited the experiences of team members, many of whom have extensive experience of working with and in embedded research environments and different levels and sections of the health system over a number of years, and whose engagement in the VAPAR process has allowed for insights into the receptivity of the system to research evidence as well as insider/outsider perspectives on systems functioning across all the health system blocks.

We also draw on public documents such as provincial and district plans, expenditure reports, media reports, other relevant evidence generated by the MRC/Wits-Agincourt Unit and relevant recent reports on the health system in Mpumalanga produced by other organisations; these were sought using key words relating to the system blocks and levels, combined with locality terms, and selected purposively to inform elements of our framework, alongside wider global and health system research from South Africa. Data or insights were drawn from 78 documents, of which eight related to the VAPAR project, 12 provided evidence on Mpumalanga province, 42 related to the wider health system in South Africa and 16 discussed decision space in other settings and conceptually.

Data analysis

Analysis was undertaken using the decision space framework and integrating data from the main strands by the research team (Table 1), initially in a research team workshop in 2019, and subsequently through iterations of the paper and subsequent reflective research team meetings. The research team conducted a rapid literature review on decision space in health care and identified a framework [5] to structure its analysis. Literature was shared with team members and discussion held on its application to the local context. Data were factual (commonly from official sources, such as provincial reports) or interpretative (thus representing important views of sectoral performance), so quality assessment of these sources was not applied or relevant. Analysis was abductive, focused on patterns within domains of the conceptual framework, comparing across levels and reaching consensus by team discussions and drafting. As decision space is not observed directly, but is emergent from the domains assessed, team interpretation was applied to derive a commentary on decision space.

Download:

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

Table 1. Summary of data sources.

https://doi.org/10.1371/journal.pone.0304775.t001

Ethical considerations

This article draws from the PAR data. For the PAR, informed consent was sought from all participants, all identifiable data were anonymised, and approvals were obtained from the authors’ institutes and from the provincial health authority. Additional information regarding the ethical, cultural, and scientific considerations specific to inclusivity in global research is included in the Supporting Information (S1 Checklist).

Facility level

At facility level, we find that decision space exists, though it is constrained by lack of resources and a culture of bureaucratic, often fragmented, upward accountability. Decision space is demonstrated by facilities which use positive informal coping strategies (such as sharing of supplies across facilities to overcome shortages) and evidenced by the range of performance by clinics in similar settings in terms of waiting times, for example.

Authority

Key roles are clearly defined in facilities. These include, in hospitals, the Chief Executive Officer (CEO) and a Management team, including a Clinical Manager and Nursing Manager, who are in-charge in hospitals, alongside (mainly in tertiary and regional hospitals) specialist staff dealing with child health, such as hospital-based paediatricians. Within clinics, Operational Managers (usually professional nurses) are in-charge.

Accountability

At hospital level, the CEO and hospital management team report to hospital boards, which oversee the work of the facility. Hospital boards have power (including to go to provincial leadership with problems) but tend not to use these mechanisms and are often focused on political priorities (TO). Although hospital boards do receive financial reports, very few take responsibility for hospital financial matters. As a result, Boards can lack capacity (see below). The Portfolio Committee report 2017/18 reports constraints to functionality in some hospital boards and that community protests have resulted in the removal of some CEOs, indicating a form of informal accountability claimed by communities [34].

Litigation is becoming more common as a strategy by families (and lawyers) to hold the DoH to account for failures, though this mainly has consequences for provincial budgets more than for individual facilities or practitioners. In 2017/18, there were 61 legal cases outstanding, mostly relating to maternal deaths [34]. At a clinical level, audits and adverse event reporting and committees are part of the quality assurance process, although these committees vary in degrees of activity and effectiveness (TO).

At primary level, community health centre or clinic committees are the main formal accountability structure. Committees should be made up of elected community representatives and health professionals to allow community concerns to be heard and addressed. Various mechanisms are prescribed to support engagement with and feedback from communities, including complaints systems, satisfaction surveys and waiting times reporting, but these are not seen as comprehensive. In addition, clinics are required to hold Open Days as part of the Ideal Clinic initiative, however from our participant observation these have limited attendance and substance, mainly as communities are generally already aware of services rendered by clinics. While home-based carers, community health workers, and Ward-based PHC Outreach Teams are visible links between communities and facilities, they face many demands, limited recognition in the system, and no additional resources to support a rapidly expanding mandate. In this context, clients tend to use direct action such as protests or the media (including increasingly social media) rather than formal channels to address grievances. These are problematic, imposing system costs and do not represent balanced or constructive input from the community.

In terms of vertical accountability, facilities have to organise services within the policies passed down to them for each programme, such as child health. VAPAR participants perceive policies, programmes, audits and other initiatives prescribed from higher levels as well-intentioned but insufficiently tailored to local conditions and resources. Furthermore, many parallel initiatives were described as having an overall destabilising effect on over-burdened health workers, compounding problems in an already constrained system. As a result, objectives are often not achieved and results are manipulated to show progress. This was described as ‘changing numbers rather than really affecting change’ [42]. Clinics are cost centres and can plan and manage day to day operations, however budgets are set at higher levels and major inputs (drugs, supplies, staffing, equipment) are procured and supplied by higher system levels. There is no longer a direct supervisory link between district hospitals and the Community Health Centres or clinics, which are supervised by PHC supervisors (professional nurses), based in the sub-district.

Capacities

Accountability is closely linked to capacities, which include resources of various kinds as well as their management. Reflecting on challenges in relation to under-five mortality, VAPAR participants highlighted insufficient and absent health workers, deficits in competencies (e.g. gaps in Integrated Management of Childhood Illness training, or in management of severely malnourished children and health education), support and motivation as contributory factors in under-5 deaths, and which collectively limit quality and foster poor attitudes more generally [42]. Responding to shortages, some staff act outside their mandated professional scope. Examples of strong personal commitment are seen in some cases, this is contrasted by poor management of patients in other cases [43–45]

A recent assessment in nine clinics in the study area identified issues with staff vacancies, with 55% reporting one or more vacancy, and some services affected by staff action over remuneration [46]. Nevertheless, in the same study, providers were highly satisfied with the nature of their work and generally with relationships with colleagues, but reported dissatisfaction with working conditions in terms of staffing, supplies and space. Staff also raised concerns about frequent absences of colleagues, lack of support staff, poor motivation, as well as infrequent and demotivating supervision and trainings which distract from clinical care. Additionally, some providers were less satisfied with opportunities for and quality of training and supervision and an average of 42% of providers (8% to 81% by clinic) reported plans to leave in the next two years.

Inadequate service organisation and infrastructure were further challenges identified by VAPAR participants. Specific issues included overcrowding, especially during morning clinics, exacerbating staffing problems and causing long waiting times and delayed or postponed consultations. Lack of consulting rooms and water outages were also cited: many clinic buildings date back to when the area was a homeland (areas designated under the Apartheid regime for African self-government) and significantly smaller populations were provided with limited and poor quality services. Public facilities now cater for much larger populations and are competing for infrastructure budget with other projects, resulting in higher throughput than is manageable [39].

The clinic quality assessment also identified long waiting times (e.g. median wait of 122 minutes for antenatal care, for a consultation lasting around 8 minutes on average [46]. However, there were wide variations by clinic: the best-performing facility had only 16% waiting over 2 hours, while the worst had half of all patients waiting over 2 hours. These data highlight significant potential in local management and organization to improve waiting times for patients.

Maintenance failures and a lack of maintenance planning exacerbate this situation. Facility maintenance is not within the control of Mpumalanga DoH and involves the Department of Public Works, Roads and Transport at provincial level. The centralised health budget makes even minor upkeep expenses difficult to access [47]. The clinic quality assessment supports these infrastructure concerns. For example, functional internet was absent in 88% of facilities, electronic medical records were absent in 78%, and functional fans were absent in 55% [46]. Patient scores were generally high but, reflecting these issues, were low for cleanliness in particular.

Shortages of ambulances and interrupted supplies of medicine are a further critical challenge [48] reflecting the lack of autonomy on purchasing, while the Portfolio Committee 2017/18 also highlights challenges in relation to infrastructure, drug supplies, equipment, clinical staffing and referrals [34]. These issues also affect Ward-based PHC Outreach Teams work as nurses cannot reach communities to support Community Health Workers in the absence of transport [42]. Reflecting the issues described above, only 8% of clinics in Ehlanzeni met Ideal Clinic standards in 2017/18, compared to a national average of 43.5% [49].

For tracer medicines, 94% of clinics had more than 90% availability, slightly above the national average. This may reflect informal coping strategies to manage shortfalls, such as: sharing supplies across clinics; shortening the length of prescriptions; or ordering more supplies than may be warranted based on patient load owing to regularly receiving less than is ordered [46].

Finally, there are challenges relating to the quality of the District Health Information System (DHIS). While the system is widely used and tracks facility data, including child deaths, limited information is provided on cause of death and there are concerns about completeness, as well as lack of use of evidence for decision making [31];. The Portfolio Committee report for 2017/18 [34] notes that patient file management in hospitals and clinics remains a problem and that responsibility for implementation of the Health Management Information System has been moved from provincial to National DoH. A National Health Patient Registration System and web-based DHIS-2 are being rolled out [34].

District and sub-district levels

The district and sub-district have an important role in operationalising policy, but vertical accountability can be punitive, rather than supportive, and resource constraints are commonly cited. However, good personal relations and communication can ease day to day functioning, enabling decision space to be opened. As an example, during COVID, the team observed local intersectoral action being taken in response to the crisis by district and sub-district health and other actors.

Accountability

In the district, there is clear upward accountability through quarterly review meetings and annual reports against district health plan targets, though feedback from above tends to be focused on problems, more than identifying, understanding, supporting and enabling local leadership, supervision and innovation (such as the coping strategies around medication shortages described above). The main accountability of the MCWYH&N programme coordinator is to the District Health Management Team, rather than to technical leads at provincial level. Downward accountability links to facilities are limited, and vertical communication often depends on personal relationships. In work to date at district level, the VAPAR learning platform was seen as valuable in terms of provision of opportunities for improved vertical communication between district and provincial levels, which was identified as a major gap, as well as encouraging constructive dialogue on problems and response strategies [43].

Capacities

Staff at district and sub-district level direct priorities for facilities but in a consultative way, for example in planning for child health campaigns. At district level, annual planning reflects divergent themes of optimism about achieving set targets while acknowledging the reality of an often under-resourced and partially dysfunctional system. Human resources challenges are highlighted in the 2018 Ehlanzeni district health plan. These include staff shortages (including support staff), high staff turnover, absenteeism, poor alignment with organisational structure and lack of outreach [47]. Hiring was decentralised to district level but then recentralised due to irregular appointments being made by district managers and CEOs. This has, however, created operational issues at district level, with long delays hiring even basic support staff (TO).

Austerity measures have been implemented annually since 2012 to improve efficiency and curb expenses. Formal and informal centralisation of procurement of goods and human resource functions has effectively limited capacity among district officials to perform duties (e.g. travelling, telephone and cell phone costs, accommodation when travelling), which is reflected in operational plans [47]. No allocation of equitable share (the funds which are send to the provinces without earmarking from national Treasury) towards goods and services was made to programmes at district and sub-district level, and with decreasing allocations at provincial level [51]. Vital equipment is reported as unavailable in some facilities, especially PHC facilities, due to budgetary constraints. Infrastructure maintenance has been failing due to centralisation of the maintenance budget, which is hard to access. Support services, such as supply chain management are slow and the competency of staff is questioned; the net effect is delays in procurement.

The sub-district has insufficient and poorly distributed community health centres (there are four when there should be 12 to serve a population of this size). Operational limitations include shortened service hours in some districts due to staff shortages, lack of transportation, poor infrastructure, lack of maintenance and ageing equipment that is insufficient in number and poorly distributed. These are all seen as contributing to a skewed utilisation of health services between PHC and secondary levels (48). Staff also report concerns over of lack of safety, both in terms of personal protection against attacks but also related to poor working environments [52].

Provincial situation

We find that decision space at provincial level is determined by personal and relational factors, more than formal authority, accountability or capacity, which poses a risk to the performance of the health system.

Authority

As reported at lower levels, roles and responsibilities are clearly and formally defined and widely understood. However, real and stated staffing do not always align and organograms often remain in draft for extended period and/or are outdated [34]. Nationwide, vacancies exist in programmes and posts for extended periods, and staff are widely called upon to informally fill vacant positions. Acting roles are often taken on in addition to formal roles and without delegated responsibility or remuneration [53]. The situation can contribute to authority vacuums, where staff do not feel empowered to take required decisions. At the same time, however, overstaffing of some senior management positions occurs, to the detriment of filling service delivery posts. Fluidity in roles and responsibilities is reported as a result, as well as insecurity for staff, waste and instability, with post holders’ status influenced by personal networks (TO).

Accountability

Formal structures exist for planning, budget-setting and performance targets in the Annual Performance Plan (APP) and for enforcing collective responsibility, as laid down in the Public Finance Management Act (PFMA) 1999 [54]. This sets out national standards, including the need for financial statements to be audited and made public by the national Auditor General [55]. The national Department of Planning, Monitoring and Evaluation provides cross-sectoral oversight, reporting to the President. There are a range of structures and processes in provinces: annual reviews, oversight by the Standing Committee on Public Accounts, and the Portfolio Committee on Health and Social Development under the Office of the Premier that are reported to have the power to call officials and department to account if key targets are missed [56].

Nevertheless, upward accountability to national level (or technical support from it) is seen as limited, given that health service delivery is constitutionally the responsibility of provinces, unless programmes are recipients of a national conditional grant—which only applies to HIV/AIDS, sexually transmitted infections and TB, or issues that are in the media for some reason. Other programmes rely on annual distribution according to the provincial APP from ‘equitable funds’, which are allocated from the national level to the province.

There is divergence between theory and practice regarding downward linkages and accountability. While planning is designed to be bottom-up, with district plans feeding into provincial plans (59), the reverse is observed to be the case in practice. Provinces generally set priorities (including for maternal and child health) with targets divided between districts. As a result, districts lack ownership of targets and accountability is limited as performance differences are lost in aggregated provincial reporting (as long as average targets are met, individual variation is not necessarily probed) (TO). Equally, budgets are supposed to be set bottom up, according to plans, but in reality they are set top-down with managers given ceilings to work within (TO).

Moreover, programme budgets can be reallocated to different and shifting priorities (even politically driven) arising in-year, with better-connected managers often relatively protected from such variance. This undermines both performance and accountability. While there are mechanisms for enforcing collective accountability, as highlighted above, individual accountability is less strong. The performance management and development system in the province is not functioning as intended, which means that, at an individual level, rewards for merit and sanctions for poor performance are selective (TO). Resources do exist in the system in the form of motivated staff, committed to public service, however they are often not supported. Similarly, sanctions are perceived as limited. For example, staff who are suspended pending investigation of malpractice have been reported to continue to draw salaries and benefit from housing over considerable periods, while others are moved to new positions, without the complaints being resolved [57].

The Auditor General’s 2013/14 PFMA report [53] on health in the province reflects these issues, citing a lack of consequences for poor management or transgressions against the Act, poor response in addressing the root causes of poor audit outcomes, and an overall lack of key controls.

Disconnects between lower levels realities and higher level policies are widely documented in the country over sustained periods and in areas not limited to child health [58]. The VAPAR process has also identified hierarchical governance as failing to account for significant local innovation, responsiveness and resilience at lower levels. Provincial participants in the VAPAR process were positive about cooperative learning processes that enable and encourage bottom-up, appreciative and reality-based learning and exchange [42].

Capacities

Mpumalanga had growth in PHC expenditure per capita in 2017/18, but is the third lowest spending province at R1,011 per capita; R144 below the national average [49]. Misappropriation was reported to play a role by the National Treasury, which reported fiscal risks and high accruals. Improved infrastructure, capacity enhancement, and better planning and project execution are sought to reduce under-spending, especially of conditional grants. A recent report also refers to irregular expenditure and medico-legal claims in health–litigation against the state in relation to health services, which are a growing (but unbudgeted) expenditure [34].

Annual budget increases are not aligned with population growth, extended service delivery priorities or policy directives. Priority initiatives such as Ideal Clinics, for example, have not come with additional funding. Moreover, unequal expenditure per population is reported across districts and sub-districts, highlighting the need to review budget allocations to reflect equity [51]. Budgets, which are key to enabling action, are frequently reallocated towards political priorities, such as building projects, which, without adjusted performance targets, frustrates service delivery.

Regarding staffing, appointments are often not seen to be based on merit or expertise, but rather on personal networks and years in service (TO). Trade unions also play a major role in the appointments process. These contribute to competency shortfalls, undermining managers’ decision space. Organograms are highly contested and while there are estimates of required workforce, these are not matched by funding or by actual distribution of staff [51].

There have been hiring freezes [47], with brief moratoria during which there are hectic attempts to fill posts. Gaps remain, especially in previously disadvantaged areas, which is also true for facility distribution. Problems with staff morale and capacity are noted as contributing to the rise in medical litigation (mostly at hospitals and often in relation to maternity care) [34].

Although vacancy rates in critical occupations (as defined in the APP, relating to essential services) were only around 8%, there are significant challenges, given misalignment of official posts, those actually filled and those who are on payroll [30, 51]. Specialist doctors at referral hospitals were particularly lacking. 6% of employees left employment in the year covered by that report, of which a large proportion (32%) was resignations. More specifically, provincial health management is described as existing in a state of ‘crisis control’, with, for example, dates for meetings set but not followed when something more urgent occurs [59].

Information flows are based on the DHIS and periodic surveys, but gaps exist, notably from the community level. In Mpumalanga specifically, a provincial health research office was established in 2016, maintaining a database of research in the province [34]. While it does not as yet actively guide or coordinate research, there are intentions to develop more input to and control over research in the province, and drawing on the VAPAR process where necessary and relevant. Mpumalanga has a relatively new university, which intends to address key health needs in future, and the province also hosts the MRC/Wits Rural Public Health and Health Transitions Research Unit (of which the HDSS is a core element)

In relation to supplies, some tenders are overseen by the Provincial Treasury, with the ‘end users’ in each provincial Department responsible for drafting specifications, as well as evaluating the bids received, which can be a source of distortion, with reports of mark-ups on the cost of goods and shortfalls in provision across the provincial health system, as well as poor quality goods [47].

Reflecting the issues described above, the Auditor General’s PFMA report 2017/18 was qualified, with findings, which had been the case for the past five years [60]. Citing direct consequences on service organisation and delivery, it reported R310 million of irregular expenditure in 2017/18, with mutually-reinforcing capacity deficits:

“Health, with the second largest budget, had significant findings relating to poor storage and stock management practices, staff shortages, insufficient training, and medical equipment that was not in a good working condition. These issues contributed to the poor health services in the province. The main drivers of the shortcomings were poor project management together with staff vacancies and instability.”

Study limitations

A major limitation of the paper is that the examination of decision space was not planned in advance but arose from an observation of its importance in influencing how evidence co-created by VAPAR partners could be put into use. The analysis was therefore conducted retrospectively and limited by available data. As a result, the paper is not able to assess decision space (in relation to different health system functions) or its impact on performance. However, we are able to observe inter-relationships between components which influence decision-space and comment on how constraints to these might be mitigated and opportunities taken to improve them.