A qualitative study of the dissemination and diffusion of innovations: Bottom up experiences of senior managers in three health districts in South Africa

Background: In 2012 the South African National Department of Health (SA NDoH) set out, using a top down process, to implement several innovations in eleven health districts in order to test reforms to strengthen the district health system. The process of disseminating innovations began in 2012 and senior health managers in districts were expected to drive implementation. The research explored, from a bottom up perspective, how efforts by the National government to disseminate and diffuse innovations were experienced by district level senior managers and why some dissemination efforts were more enabling than others. Methods: A multiple case study design comprising three cases was conducted. Data collection in 2012 – early 2014 included 38 interviews with provincial and district level managers as well as non- participant observation of meetings. The Greenhalgh et al. (Milbank Q 82(4):581-629, 2004) diffusion of innovations model was used to interpret dissemination and diffusion in the districts. Results: Managers valued the national Minister of Health’s role as a champion in disseminating innovations via a road show and his personal participation in an induction programme for new hospital managers. The identification of a site coordinator in each pilot site was valued as this coordinator served as a central point of connection between networks up the hierarchy and horizontally in the district. Managers leveraged their own existing social networks in the districts and created synergies between new ideas and existing working practices to enable adoption by their staff. Managers also wanted to be part of processes that decide what should be strengthened in their districts and want clarity on: (1) the benefits of new innovations (2) total funding they will receive (3) their specific role in implementation and (4) the range of stakeholders involved. Conclusion: Those driving reform processes from ‘the top’ must remember to develop well planned dissemination strategies that give lower-level managers relevant information and, as part of those strategies, provide ongoing opportunities for bottom up input into key decisions and processes. Managers in districts must be recognised as leaders of change, not only as implementers who are at the receiving end of dissemination strategies from those at the top. They are integral intermediaries between those at the at the coal face and national policies, managing long chains of dissemination and natural (often unpredictable) diffusion. The research specifically sought to understand, from a bottom up perspective, how efforts by the National government to disseminate and diffuse NHI piloting innovations were experienced by district level senior managers, in order to draw key lessons from the experience on practices that enable the diffusion and dissemination of innovations in ways that are helpful for health managers at the coal face. In 1994, the South African government inherited a fragmented and regressive health system that was set up to serve a minority of the population with a hospital based curative focus [28]. Embodying a vision to establish an equitable health system with a PHC focus, the government developed a National Health Plan for South Africa in 1994 which laid the foundation for establishment of a district health system [29]. South Africa has a fiscal federal system of government which decentralises authority for a range of powers, functions and budgeting from National to Provincial and Local government. National government is the main revenue collector and financial transfers are made via equitable share allocations (general-purpose) and conditional grants (specific purpose) to provincial and local governments to provide and finance services in the different sectors (including health) [30]. The National Department of Health (SA NDoH) is primarily responsible for policy making, setting standards and regulations while the Provincial Departments of Health are responsible for the provision and financing of public health care services relying largely on the National government for financial resources. Local municipalities are responsible for environmental health services, and PHC services are delivered through the district health system (DHS), which is by design a lower level of the provincial health authority [30]. The Provincial government determines the amount of decision space granted to each DHS, including financial and human resource authority. The DHS is led by a district manager (DM) who works together with programme, hospital and support service managers as a district management team; across the fifty-three health districts, the structure of district management teams vary in practice [31]. In South Africa, management competence in facilities and districts at all levels is varied with managers having a diversity of backgrounds [32]. Today South Africa has fifty-three health districts established through the 2003 National Health Act. The district health system is still being institutionalised in many ways and whilst it has achieved gains, the current challenges include different interpretations by Provinces on what constitutes the most appropriate structure for a DHS, sufficient delegation of powers to managers in districts, that formal accountability mechanisms are not always in place, shortages of human resources and district hospitals that are sometimes poorly coordinated with PHC services [33]. Context-sensitive and flexible approaches are needed to understand and support evaluation of large scale health programmes due to the longitudinal and complex nature of implementation [34]. We employed a case study design, appropriate for research of contemporary phenomena in complex health systems where events and experiences of change emerge while research is undertaken, the phenomena being directly influenced by the context [35]. More specifically, a multiple case study design allowed for deeper explanation of the experience through more than one case [36]. We defined the case as the experience of district level senior managers of the dissemination and diffusion of innovations in the period 2012 – early 2014. We explored this case in three district sites, each district thus serving as a case study. Three sites were selected from the 11 NHI pilot districts. Selection criteria included (1) a district that was actively receiving information from other levels of government and/or was implementing some of the innovations, (2) access to the site, meaning district managers were prepared to give us access to staff and (3) rural/urban mix to capture variation in experiences of implementation possibly linked to geography. To maintain anonymity, only general information about the study site is given to provide context. The sites were underperforming relative to other districts in the country. For example, in 2013/14, the sites had facility maternal mortality ratios and an incidence of under 5 severe acute malnutrition higher than the national average [37]. The sites are nested in geographic areas where health and social services were hugely under-resourced and neglected during the apartheid era pre-1994 [38]. As a first step, we conducted interviews in late 2012 – early 2013 with 7 provincial level managers who played a role in information sharing and/or rolling out NHI piloting, to help purposively select the three district study sites. Guides included questions relevant to early dissemination efforts between the three levels of government, leadership for NHI piloting and information on roll out to the districts. While the paper focuses on the experiences of senior managers at district level, information from provincial interviews were relevant to contextualizing experiences and were thus included in analysis. Between September 2013 and July 2014, we undertook 2 site visits to each district, interviewing 31 members of DHMTs. Participant selection criteria included (1) being a member of the DMT (a senior manager) and (2) involvement in the dissemination of and/or implementation of NHI piloting innovations in the district. We also attended meetings as non-participant observers. Recruitment started when we presented the research protocol to the district manager in each site for approval, needed to conduct the research. At this time, we also requested the district manager to identify management team members who met the selection criteria as interviewees. In interviews with these managers we also asked them for their ideas about prospective participants to reduce selection bias. Each prospective participant was e-mailed an information letter about the project as well as a consent form and both were discussed before the interview. The semi-structured interview guide included questions that would allow us to explore the respondents’ experience of the process, including personal understandings of the vision and goals of NHI piloting, key activities taking place in the district around NHI piloting (including early communication), key activities and assumptions driving the dissemination of these activities as well as individual feelings about involvement in the process of change (individual roles, responsibilities and relationships with others) – all from the perspective of the managers themselves, not from the policy documents. Since multiple theories of change may co-exist in processes of reform, we included questions seeking to elicit assumptions and gather information on NHI piloting from the perspective of managers. We also included prompts in the interview guide from the Greenhalgh et al. [21] dissemination strategies (e.g. the role of networks, champions etc.) to help identify dissemination strategies and any emergent diffusion processes that managers were exposed to. The use of theory and thick description supports the transferability of lessons beyond the cases [36]. Cross case analysis helps to deepen understanding and explanation beyond that which a single case study can provide [36]. Cross case analysis allowed the patterns and underlying explanations to be compared, supporting transferability across sites [36]. In each site, all interviews were transcribed verbatim. The first author led the process of analysis. The first author developed a deductive coding matrix in table format, using the strategies for diffusion and dissemination identified in the Greenhalgh model [21] as headings to support data extraction and analysis across the three sites [36]. A code book was developed to ensure each researcher understood each deductive code. Allowance was made in the deductive table for inductive coding to capture any emergent findings and ideas. In each site, data were manually extracted from each interview into the deductive coding matrix by a site research team. The completed matrix from each site was reviewed by the first author to identify key themes and explanations related to manager experiences of diffusion and dissemination in each site. For each site an initial story of what factors were enabling or constraining national government efforts was developed, this was done in consultation with site research teams to promote rigour in the analysis. The lead author then identified key similarities and differences across the three sites for inductively developed themes that helped to answer our research question. The lead author, in a more deductive approach, also drew on the Greenhalgh et al. [21] strategies to look for patterns in each site, specifically grouping codes into named patterns of ‘strategies and actors related to dissemination’ and ‘strategies and actors related to diffusion’, as well as looking for any factors related to the context (innovations system fit) that may have enabled (or not) the diffusion and dissemination of innovations in that site. These patterns were discussed and verified with the research team. Parts of this research was also fed back to managers from the three sites in a one-day feedback session, this provided some opportunity for member checking. The findings of this paper will be developed in to a policy brief as well as incorporated into our teaching, which includes many students working in the health system in South Africa. Draft cross case findings were written up by the first author and reviewed in iterative rounds by researchers across the three sites until a final synthesis was reached. The process of writing was also a source of rigour as co-authors were able to verify the lead author’s synthesis as it evolved. The first author also reflected on top down and bottom up implementation theory [39] to situate the findings within the broader government context within which dissemination and diffusion was taking place.