Multisectoral (in)action: towards effective mainstreaming of HIV in public sector departments in South Africa

Whilst progress has been made, evidence of effective approaches to improve action on addressing the social and structural drivers of the HIV epidemic remains a priority, to meet the 2030 sustainable development agenda, and to achieve key HIV targets, including the 90-90-90 target and the Treatment as Prevention (TasP) intervention. With a focus on the public sector in South Africa, we critically reflect on the HIV mainstreaming approach, assessing its ability to augment multisectoral action on the response to HIV. We reflect on progress made in mainstreaming HIV in non-health sector departments, exploring factors that have enabled and hindered the process. We also highlight limitations in the adopted approach to mainstreaming HIV in non-health sector departments in South Africa; which currently promotes working in silos and does not encourage collaboration and partnerships. We propose a three-step approach to effective mainstreaming of HIV that will augment multisectoral action. The approach also contributes towards realising the sustainable development agenda of “leaving no one behind” and achieving the national and global targets on HIV that are embedded in collaborative efforts.


Introduction
Sub-Saharan Africa remains home to the largest number of people living with HIV globally (UNAIDS, 2017b). An estimated 7.9 million people were living with HIV in 2017 in South Africa (HSRC, 2018;Republic of South Africa, 2017). Whilst progress has been made, evidence of effective approaches to improve action on addressing the HIV epidemic is urgently needed. In 2018, pressure to make progress on the HIV and AIDS response remains a key priority in order to meet the 2030 Sustainable Development Goals (SDGs) agenda of "leaving no-one behind", and to achieve key HIV and AIDS targets, including the 90-90-90, and the goal of Treatment as Prevention (TasP) (UNAIDS, 2017a;WHO, 2012). The HIV and AIDS response provides a platform for action on which to build on to achieve the SDGs and new priorities on prevention as treatment, and to strengthen efforts to address health in an integrated way (Sedibe, Nygren-Krug, McBride, & Buse, 2018;UNAIDS, 2015).
Social and structural drivers increase risk to acquisition of HIV, and vulnerability to the impact of AIDS; these are multifaceted and cannot be addressed by one sector (Auerbach, Parkhurst, & Cáceres, 2011;Commission on Social Determinants of Health, 2008;Seeley et al., 2012). Multisectoral action is recognised as central to addressing the social and structural drivers of the epidemic, and a key strategy to the target of ending HIV in 2030 (Seeley et al., 2012;Spicer et al., 2010;UNAIDS, 2015). However, despite this recognition, effective multisectoral responses are lacking and evidence exists to suggest that this is a gap in the response to HIV (Gavian, Galaty, & Kombe, 2006;Hellevik, 2012;Hemrich & Topouzis, 2000). A declaration was signed at the United Nations General Assembly Special Session (UNGASS) in 2001, by countries including South Africa, committing to implement multisectoral strategies at national level. The 2001 UNGASS Declaration enjoins countries to integrate HIV and AIDS responses in planning frameworks, and budget allocations of all development sectors (UNAIDS, 2001(UNAIDS, , 2004UNDP, 2005). To achieve this, countries are expected to engage in a process of mainstreaming HIV, and address the epidemic in an integrated manner (UNAIDS, 2001;UNDP, 2005).
HIV mainstreaming is considered an appropriate and sustainable strategy to enable a multisectoral and multistakeholder response (UNAIDS, 2004;UNDP, 2005). Mainstreaming is defined as a process that enables development actors -including both state and non-state actors to address the causes and effects of HIV and AIDS in an effective and sustained manner, both through their usual work and within their workplace (Republic of South Africa, 2011a;UNAIDS, 2004;UNDP, 2005). Working together, sectors would analyse the impact of HIV and AIDS, both internally (within the workplace) and externally (in the communities where they work), determine the response based on their comparative advantage, and accelerate national responses to the epidemic (UNAIDS, 2004;UNDP, 2005). Through HIV mainstreaming, sectors would commit to specific activities to achieve joint outcomes for the national

Multisectoral (in)action: towards effective mainstreaming of HIV in public sector departments in South Africa
HIV and AIDS response, and ultimately contribute towards achievement of global HIV targets (UNDP, 2005).
While the concept of mainstreaming has been with us for over a decade, its application and implementation in practice continues to evolve (Dube, Dube, Mpofu, Magure, & Taramusi, 2016). In relation to this, other mainstreaming initiatives in areas such as nutrition, disability and gender have been documented, and continue to inform the process of mainstreaming HIV (Bhandari, Kabir, & Salam, 2008;Mannell, 2010;Nyamhanga, Frumence, & Simba, 2017;Pelletier et al., 2012;Sweeney & Riddell, 2014). However, to date, there has been limited reflection on if and how mainstreaming contributes to an effective multisectoral HIV response that addresses the structural drivers of the epidemic. The current South African National Strategic Plan (NSP) on HIV andAIDS (2017-2022;Republic of South Africa, 2017) and the two previous NSPs -2016Republic of South Africa, 2012) and (2007Republic of South Africa, 2007b) call for mainstreaming HIV and AIDS across all government departments -including health and non-health sectors. Despite this, for many years HIV in practice has been viewed as a biomedical issue, that only concerns -and is the responsibility of -the health sector (Republic of South Africa, 2007b. Based on both anecdotal and empirical evidence, mainstreaming of HIV in South Africa has, since its inclusion in the 2007-2011 NSP (Republic of South Africa, 2007b), become a buzzword, often mentioned but rarely defined or critiqued, with increased expectation for this undefined approach to augment a multisectoral response to HIV that involves other sectors beyond the health sector.
To reflect on South Africa's experience in HIV mainstreaming, and the extent to which the mainstreaming approach augments multisectoral action in practice, this paper draws on research with government departments to assess progress and to understand the factors that both facilitate and hinder the mainstreaming of HIV in non-health sectors. The paper also explores how HIV mainstreaming is coordinated, vertically between levels of government (national, provincial and local) within departments, and horizontally between departments. This is done to critique the mainstreaming approach, and its contribution (or not) in augmenting multisectoral action in the response to HIV. This paper focuses on mainstreaming of HIV in the public sector in South Africa. Involvement, of civil society and the private sector in the multisectoral response to HIV in South Africa is reported elsewhere (Mahlangu, Vearey, Thomas, & Goudge, 2017).

The structure of government in South Africa
South Africa is constituted on a quasi-federal system with three spheres of government at national, provincial and local levels (Muthien, 2014;Schneider & Stein, 2001). Government is structured such that all levels of government are expected to be involved in the HIV response in different capacities (Lodge, 2015). National government is responsible for developing broad policy frameworks, defining norms and standards for service provision, and distributing revenue in an equitable manner; provincial and local governments are responsible for implementing most public functions and for putting into effect the policies developed at national level (Republic of South Africa, 2003;Schneider & Stein, 2001).
The 1996 Constitution provides that the three spheres of government should function in a cooperative, interrelated and interdependent manner, and ensure that there are intergovernmental relations, formal and informal processes, and institutional arrangements and structures for coordination within and between spheres of government (Republic of South Africa, 1998). Different structures and processes exist for coordination -"a process to ensure that activities and functions of the three spheres of government do not overlap and that functions are not duplicated"between the three spheres of government (Malan, 2005). Table 1 describes the systems and structures for intergovernmental relations in South Africa, which are general, and not specific to HIV. Coordination involves recognised interdependence of two or more organizations coming together to address a common problem, through reconciliation of different processes and activities, which take place simultaneously and consecutively (Hogl, 2002).
The literature describes two forms of coordination: horizontal coordination between organisations on the same level; and vertical coordination between organisations at different levels of government (i.e., regional, national or local) (Christensen & Laegreid, 2008;Perri 6, 2004;Pollitt, 2003).

Historical developments that have shaped the public sector response to HIV over time
The development of the public sector response to HIV and AIDS in South Africa has been slow and complicated, characterized by tensions and overlaps in roles and responsibilities between spheres of government, and in some cases lack of clarity in weaving the response to align with the functions and a mandate of some sectors, which  (Schneider & Stein, 2001;Schutte, Weck, & Boessenkool, 2004). The South African National AIDS Council (SANAC) was established in 2000 to function as a multisectoral coordinating structure, opening space for other government departments, civil society and the private sector to participate in the response to HIV (Republic of South Africa, 2000). There was recognition that HIV had become a health and development issue, and that the narrow biomedical response was no longer adequate. However, the location of SANAC within the National Department of Health (NDoH) meant that HIV and AIDS was still perceived by some departments as a health issue, which was outside their mandate (Wouters, Van Rensburg, & Meulemans, 2010). Even departments that were involved at the time, mainly focused on promotion of HIV campaigns in the workplace, with limited understanding of what role they needed to play (Schneider & Fassin, 2002). There were also divisions between spheres of government, with some provinces openly defying the national government by initiating antiretroviral treatment (ART) to HIV-positive pregnant women, the national government was reluctant to roll out the drug, questioning its safety and affordability (Fourie & Meyer, 2016). Premised on this, the former President Thabo Mbeki publicly questioned the causal link between HIV and AIDS, believed that HIV is harmless and that AIDS symptoms are caused by malnutrition, drug abuse and even antiretrovirals themselves. Similarly, his Minister of Health, Manto Tshabalala-Msimang, supported untested natural therapies (beetroot and garlic) for treating HIV, rather than ART (Nattrass, 2008).
A turnaround in political commitment, and in the response to HIV was depicted when the newly elected political leadership at the time (2009), publicly acknowledged HIV as among the most important challenges facing the country; and the newly appointed Minister of Health brought urgency and a renewed focus to the HIV response (Simelela, Venter, Pillay, & Barron, 2015;Wouters et al., 2010). The call for mainstreaming HIV in government departments was formally incorporated in the second NSP (2007-2011. For the first time, all government departments were expected to develop mainstreamed HIV plans and contribute towards achieving a coherent, countrywide response to HIV. In addition, SANAC was revived, with the mandate to oversee establishment of similar multisectoral structures to coordinate implementation of the response to HIV at provincial and local levels (Republic of South Africa, 2007b).
The third NSP (2012-2016) has restated the mainstreaming approach and multisectoral action on the HIV response. It continued to advocate that government departments should "get involved" in the response to HIV. Addressing the social and structural drivers of the epidemic was also top of the agenda in the plan, thus recognition of the role of other sectors beyond health. There was also emphasis on implementation of combination prevention approaches, and a focus on key populations including sex workers, men who have sex with men and other groups (Republic of South Africa, 2012a).
The latest NSP (2017-2022) has multisectoral action as one of the guiding principles -with an expectation that government departments should mainstream HIV, develop plans outlining how they will contribute -to achieve the goals of the NSP. Treatment as Prevention and the international developments regarding the Universal Test and Treat strategy have become top priority in shaping South Africa's response to HIV. The country is also gearing towards a vision of an HIV response based on "whole-ofgovernment" approach, where all departments would join-up, working together to address the HIV epidemic (Republic of South Africa, 2017).

Planned approach to mainstreaming HIV in the public sector in South Africa
NSPs should provide guidance to stakeholders involved in the response to HIV: define the national targets and approach, and guide the development, implementation and evaluation of national responses to HIV (Hanass-Hancock, Strode, & Grant, 2011;Wouters et al., 2010). While the call for mainstreaming HIV in government departments in South Africa was first incorporated in the NSP 2007-2011, to date SANAC has yet to develop a HIV mainstreaming framework. SANAC is the highest level coordinating body for the country's response to HIV, responsible for overseeing implementation of NSP; it is meant to help realise the call to mainstream HIV. Guidelines on gender-sensitive and rights-based mainstreaming of HIV in the public sector were developed by the Department of Public Service Administration (DPSA) and only published five years later, in 2011 (Republic of South Africa, 2011a). In 2007, the Department of Cooperative Governance (CoGTA), which was the Department of Provincial and Local Government (DPLG) at the time, developed a framework to guide mainstreaming of HIV in local government (Republic of South Africa, 2007a).
Whilst mainstreaming is meant to facilitate multisectoral action on HIV in the country, the DPSA guidelines fail to unpack this connection. As a result, the guidelines are limited to describing the process of developing HIV plans in government departments, and what should be the content or substance of the plan. The guidelines are silent on the process of coordination of HIV plans (HIV responses) between levels of government in departments (national, provincial and local), and between departments (e.g., DSD, DBE and DHS). Highlighted as a principle in the guidelines is that "strategic partnerships based upon comparative advantage, cost effectiveness and collaboration must be developed for effective implementation of HIV mainstreaming"; yet, no guidance is provided on how that will be achieved (Republic of South Africa, 2011b: p. 19).
The DPLG framework outlines the "what", "who" and "why" of HIV mainstreaming in local government. The framework advocates for HIV mainstreaming, defines the concept, contrasts HIV mainstreaming with HIV programming, and describes who the key actors in HIV mainstreaming in municipalities should be, and their roles and responsibilities. Unlike the DPSA guidelines, the DPLG framework provides more depth in that it describes how coordination between various departments within the municipality should be done, through the integrated development planning (IDP) process (Republic of South Africa, 2007a).

Study design
To reflect on the process of HIV mainstreaming in the public sector in South Africa, and to critique the value-added (or not) of the mainstreaming approach in augmenting multisectoral action on the response to HIV, this studyinitiated in 2013 -used an exploratory qualitative study design. The reflection is undertaken in light of recent developments and priorities related to the AIDS responses (globally and nationally) to "leave no one behind" and "treatment as prevention". Mpumalanga province, along with three of its districts and six local municipalities, were purposively selected as the case study for in-depth exploration of responses to HIV at the provincial and local levels. The province is one of the four provinces in South Africa with a high HIV prevalence (22.8%) among adults aged 15-49, after Kwazulu-Natal (27%), Free State (25.5%) and Eastern Cape (25.2%) provinces (HSRC, 2018).
The research involved a review of HIV policy documents; key informant interviews; and participant observations. Participant observations were conducted to understand the process of development of HIV plans in government departments. Of particular interest was to learn: which stakeholders were involved, their level of engagement and recognition during the process, and how priorities in terms of programmes were made. We also observed how AIDS council meetings were undertaken, to learn the issues discussed, and how different sectors interact during the meetings. The key informants were located in government departments and structures responsible for coordinating the HIV response at national, provincial and local levels.

Selection of government departments
Purposive sampling was used to select and include departments (sectors) whose mandate focuses on development and health, and links with some of the social and structural drivers of HIV that are prevalent in South Africa (Republic of South Africa, 2011b). After a mapping phase, five government departments were included in the study: Department of Basic Education (DBE); Department of Social Development (DSD); Department of Health (DoH); Department of Human Settlements (DHS); and Department of Cooperative Governance and Traditional Affairs (DCoGTA). Four of the five departments were non-health sectors; the Department of Health represented the health sector. Given the quasi federal system of government in South Africa, the research was structured to include national, provincial and local level within the selected government departments as informed by the organisation of the HIV and AIDS response in South Africa, shown in Figure 1. The district Mayor

LAC Chairperson
The local Mayor

GOVERNANCE & COORDINATION OVERSIGHT ROLE
HIV mainstreaming in South Africa (Bowen, 2009). See Table 2 for a list of documents reviewed. While the study was based on the NSP (2012-2016), we also included a currently adopted NSP (2017)(2018)(2019)(2020)(2021)(2022). Both NSPs call for mainstreaming of HIV in government departments, as a strategy to accelerate multisectoral action on the national response to HIV. A document review guide was developed to structure the review process. The review guide was informed by the Walt and Gilson (1994) model for health policy analysis, focusing on understanding the content of the policy document, actors, context and process of mainstreaming HIV in health and non-health sector departments. The NSP was reviewed to explore how the process of mainstreaming and coordination are framed, and the expected implementation plan. Progress reports were reviewed to reflect on the actual process of mainstreaming in health and non-health sectors, and a matrix was developed to pull-out key achievements, challenges and recommendations proposed to improve mainstreaming of HIV in government departments. HIV plans of government departments were reviewed to understand the department's: (1) history of involvement in the response to HIV and AIDS; (2) focus area/comparative advantage; (3) underlying premise of the strategy; (4) implementation plan; (5) and, if (and how) departments collaborate in their response to HIV and AIDS.

Key informant interviews
Mapping of key informants was conducted in preparation for data collection, to identify people involved in the response to HIV beyond the HIV workplace programme in departments. Key informant interviews (KII) were conducted with 20 respondents: 9 from the 5 selected government departments (at national and provincial level); 1 from SANAC; and 1 from DPSA. We only interviewed one official from the department of human settlement at national level, where we learned that the department only focuses on the workplace response to HIV, and there was no strategy on HIV mainstreaming that we could review. HIV coordinators from two of the three district municipalities, and from six local municipalities were also interviewed (local level).
The key informants were purposively selected to include government officials from the HIV and AIDS directorates/ units involved in both workplace and community response to HIV at national, provincial and local levels. Two national level coordinators (from SANAC and DPSA) were also purposively selected to draw from their understanding of the process of coordination, and their reflection on the challenges and how they could be addressed. An open-ended interview guide was used to conduct the interviews, and all the interviews were conducted in English, recorded and transcribed. The interviews in government departments reflected on: (1) how HIV and AIDS mainstreaming is understood and approached; (2) if (and how) government departments understand the impact of HIV in their work; (3) their response both in the workplace and in the community; (4) if (and how) the response is coordinated between spheres of government at national and provincial level within departments; and (5) if (and how) departments collaborate in their response to HIV and AIDS.

Participant observations
We participated in working group sessions during the three consultation workshops conducted by DBE and DSD in 2012-2013 during the development of their HIV plans. The workshops were attended by representatives from provincial departments, other government departments, academic and research institutions, civil society organisations and representatives from international organisations (including the Joint United Nations Programme on HIV/AIDS (UNAIDS)) and donor agencies (including the United Nations Agency for International Development (USAID)) to review and give inputs on the plans. We also participated in two Mpumalanga Provincial AIDS Council (MPAC) meetings, two District AIDS Council and six Local AIDS Council meetings which are platforms where government departments came together with civil society and the private sector to plan, monitor and report progress on implementation of HIV activities in their departments.

Data analysis
Data were transcribed and analysed using thematic content analysis. Both the data-driven inductive method and a deductive method drawing from the Christensen framework of coordination (Christensen & Laegreid, 2008) were used. Christensen's framework was selected because it best captures and allows for reflection on coordination of the HIV response vertically between spheres of government, and horizontally between departments. The deductive approach was used to formulate the objectives, structure the analysis, and to present the findings focusing on the two dimensions of coordination (horizontal and vertical), while allowing for the themes to emerge directly from the data using inductive coding. The transcripts were manually coded, and a codebook was developed in MS Word to summarise the data. All transcripts were read and data coded by PM. The codebook was shared with JV, and the findings from the analysis further discussed with JV and JG.

Ethical considerations
Ethical clearance (M120657) was obtained from the Human Research Ethics Committee (Medical) of the University of the Witwatersrand. All the participants gave written consent to participate in the study.

Results
Based on the themes identified during data analysis, the results are presented thematically: (1) mainstreaming of HIV; (2) coordination of the HIV response vertically; and (3) coordination of the HIV response horizontally. This paper presents and synthesises subthemes within each of the main themes.

Mainstreaming of HIV in government departments
Progress has been made in mainstreaming HIV in the public sector, and this includes in both health and non-health sector departments. Four of the five departments included in our study have developed HIV and AIDS plans, as required in the NSP and the DPSA guidelines. There is now a wide recognition and willingness to address the social and structural drivers of the epidemic in line with the department's mandate and function.
Approach and advancement in HIV mainstreaming in non-health sectors DBE approaches HIV as a developmental challenge impacting on educational outcomes (Republic of South Africa, 2012a). The department's role clarified in the interview with the respondent who explained that: …the Department's approach to HIV is founded on a social development perspective, and the response targeted at individual, interpersonal, family, and community levels as key structures which determine the wellbeing of society (DBE national KII). DCoGTA has adopted a development and governance perspective, understanding that: …development conditions heighten the risk of HIV infection and susceptibility to HIV. Local government's responsibility to HIV is linked to its developmental mandate stipulated in the White paper (1998). Local government is affected as an employer and provider of services to communities infected and affected by HIV and AIDS (Republic of South Africa, 2007a). DSD's approach to HIV and AIDS is founded on a social ecological model, which recognises that:

HIV and AIDS does not only affect individuals, but disintegrates families and societal systems, increasing the need for social development and welfare services in communities which they serve (DSD national KII).
The approach to HIV in three of the four non-health sector departments highlights the department's awareness of the impact of HIV and AIDS and recognition of their role and responsibility in the response to HIV. As such, it is not just "business as usual" in the way in which they delivered on their core mandate (services), but ensuring that their programmes and plans integrate HIV. We, however, found that the department of human settlement only focused on addressing HIV amongst employees, and there was no plan to address HIV in the community, like in other departments.

Departmental strategies and their potential contribution to the HIV response
All four departments that had HIV plans, had their plans aligned to the NSP (2012-2016), outlining how each department would contribute to meeting the goals and targets set out in the national response to HIV (NSP). There was a clear understanding of the role that each department needed to play in response to the impact of HIV, in line with their mandate: The department recognizes the protective role of education in reducing young people's vulnerability to HIV; and increasing knowledge and skills of learners, educators, school support staff and officials in dealing with the infection (Republic of South Africa, 2012b). DSD focuses on providing developmental social welfare services and psychosocial interventions to contribute to social behaviour change, and to reduce poverty, vulnerability and the impact of HIV and AIDS (Republic of South Africa, 2012c). Focusing on prevention, treatment, care and support, the department provides health services including ART, condoms, HIV Counselling and Testing, the Medical Male Circumcision, and the Prevention of Mother to Child Transmission (Republic of South Africa, 2012d). The framework aims to build capacity on HIV mainstreaming in local government to ensure that HIV is integrated in all municipal planning, and budgeting processes (Republic of South Africa, 2007a).

Factors that facilitated mainstreaming of HIV and AIDS in non-health sector departments
Data from this study indicate that the impact of HIV facilitated mainstreaming of HIV in non-health sector departments, which previously considered it as a health issue. There were two dominant areas of impact highlighted in our study.
Impact on capacity to deliver services (key function) The official from DoE explained: Day by day we see HIV affecting productivity of the sector, impacting on learning and teaching, with some learners and teachers constantly absent from school for extended periods of time, and others dying as a result of the illness (DBE national KII).
Others spoke about the spillover effects that extend beyond an individual infected by HIV, "increasing the number of people depending on state support" (DSD national KII).

Impact on the economy and on finances
Participants also spoke about the negative impact of HIV on the workforce and on the economy.
The The department's budget has increased over the years due to the impact of HIV and AIDS. Children are likely to need additional care and support because of death of a parent due to HIV and AIDS, and the department has to fill that gap (DBE national KII). It was evident in our study that the impact of HIV threatens efficient functioning of government departments, and precipitated a response to both health and non-health sector departments.

Vertical coordination of the HIV response between spheres of government in departments
Mainstreaming of HIV in departments also required coordination between spheres of government within departments, given the different roles and responsibilities at national, provincial and local levels. Certain conditions facilitated the process.

Consultative processes between spheres of government
The consultative engagements which were conducted between national and provincial departments were commended as a useful tool to involve everyone, given that the HIV plan is developed at national level, while implementation is meant to happen at provincial and local levels: Consultation workshops and interviews with provincial stakeholders were useful, they were conducted not only so we can tick a box that it is done, but to allow them [provinces] to be part of the process (DSD national KII). This view was shared in other departments: The national department together with national advisors that were involved in the development of the strategy, facilitated the consultative process with all relevant stakeholders in branches and provinces to ensure that they also participate in the department's response as implementers (DoH national KII). The engagement processes were also undertaken to clarify roles and responsibilities, provide support, and ensure that everyone is working towards a common goalcontributing to the department's response to HIV: We engaged in consultation processes at national, provincial and with various directorates within DBE to ensure that there is buy-in, and also assist teachers in their newly expected role of going beyond their usual function (teaching), and to start thinking about and integrating HIV and AIDS in their work (DBE national KII).

Communication and information sharing between spheres of government
Respondents also spoke about the importance of communication and ensuring information flow between directorates and between spheres of government as a critical enabler to the process: Communication between directorates, and between national and provincial and district offices has been one strategy we have found useful in the department… to ensure that everyone is on-board and that we all sing towards the same tune (DBE national KII). She further elaborated on the importance of timing of communication during the policy process: Establishing lines of interdepartmental communication between spheres of government should start early in the process to make sure that no one is left behind, and you realise when it is too late that all the plans you have are stalled because the information was not shared or communicated to key people who will be expected to take the process forward (DBE national KII). Participants also reflected on processes that hinder coordination between spheres of government within departments.

Working in silos
Working in silos between various directorates in departments was described as a hindrance to the process of coordination: All directorates have focal areas, and most times the other is not aware about programmes undertaken in other directorates. This limits ability to identify synergies and deliver programmes in a more efficient manner (DPLG national KII).

Reporting systems that are not synchronised
While one of the key functions of SANAC is to monitor and evaluate progress on implementation of the national response by all the sectors (civil society, government and the private sector), our research has highlighted gaps in the reporting process at national level: . There is currently lack of coordination of the public sector reporting process on the response to HIV between SANAC, DPME and DPSA.

Horizontal coordination of the HIV response between sectors
Our research showed that there was a wide recognition of the need for a "whole of government" response to HIV amongst departments included in the study. This was noted in several iterations in HIV plans of government departments, acknowledging that implementation of the NSP "requires strong intersectoral collaboration from different government departments" (Republic of South Africa, 2012c); that "the response requires effective partnership between government departments" (Republic of South Africa, 2012b); and "need for collaboration with other sectors to address structural enablers of HIV acquisition" (Republic of South Africa, 2012d).

Collaboration between departments
However, we found limited evidence of collaborations and partnerships between government departments on the HIV response in practice. Only two examples were found of HIV programmes implemented in partnership between departments: the Integrated School Health Programme (ISHP) of DSD, DoH and DBE; and a capacity building initiative on HIV mainstreaming by DCoGTA and SALGA.
…the ISHP brings government together, and can never be achieved without collaboration with other government departments directly responsible to provide services that are needed in schools, but are not a direct function of DBE (DBE national KII) Clarity on roles and responsibilities is important in collaboration between departments: DOH is responsible for provision of the package of school health services; and DBE's role includes creating an enabling environment for implementation of the programme in schools, planning, managing and monitoring programme implementation; DSD supports implementation of the programme (Republic of South Africa, 2012d). A respondent from DCoGTA also spoke about their collaboration with SALGA, aimed at building capacity in HIV mainstreaming in local government: We work closely with SALGA, not only in our department processes, but they also invite us to participate and engage in their processes. The recent example is when they took over the roll out of the Framework in the six provinces where we [DCoGTA] could no longer provide the mentorship programme, due to budget constraints and the restructuring process that happened in our department. Collaboration becomes a challenge when there are overlapping responsibilities, which makes the process seem like competition, instead of complementing one another: It gets confusing sometimes between us and SALGA since our mandates are overlapping. For example, after the framework was developed, SALGA also developed guidelines for municipalities which was confusing for municipalities not knowing whether to follow our framework or that of SALGA (DPLG national KII).

Existing platforms to facilitate collaboration between government department
In addition to the general intergovernmental relations systems and structures that have been institutionalized (described in Table 1), AIDS councils have been established at national, provincial and local levels to function as a platform for regular interaction, and collaboration on programmes between government departments (and with other sectors). At national level, the SANAC Inter-Ministerial Committee (IMC) is one such structure, made up of 22 government ministers/ deputy ministers, and chaired by the Deputy President of South Africa (SANAC, 2016). While the IMC is meant to be a platform for deliberation and coordination of efforts within government, not all ministers or deputy ministers participate, limiting its effectiveness (SANAC, 2016). The Provincial AIDS Council (PAC) is also meant to be a provincial level platform for interaction between government departments, yet a similar challenge of lack of participation was noted: Not all departments participate in the PAC, and in one of the meetings, the Premier of the province highlighted the need for a consolidated report from the government sector, which currently does not exist (DSD provincial KII). Lack of representation of government departments in AIDS councils was also noted across district and local municipalities where "not all government departments participate, most members are representatives from civil society organisations" (Umjindi HIV Coordinator KII). This renders AIDS councils ineffective as a structure for regular interaction and engagement.

Measuring impact of collaborative efforts
Ensuring that evaluation of progress on the response to HIV acknowledges multisectoral efforts of many departments that are involved was highlighted in the interviews: You

Discussion
The public sector response to the HIV epidemic has evolved over time in South Africa; from a biomedical framework driven by the health sector, to one that recognises the role and contribution of non-health sectors in addressing the social or structural drivers of the epidemic. The impact of HIV which threatened the departments' capacity to deliver on key functions, and the increase in budgets precipitated the shift. Non-health sector departments could no longer ignore the impact, and started integrating HIV in their programmes and budgets, tailoring the response in line with their core mandate and comparative advantage. The cross-cutting impact of HIV has facilitated involvement, with sectors recognising that each has a role in the response to HIV. As observed by others, framing of the problem or the issue of HIV, in our study, from a developmental perspective was a critical step in bringing in non-health sectors to be actively involved in the response to HIV (Clark, 2006;D'Angelo, Pollock, Kiernicki, & Shaw, 2013). Non-health sector departments were able to locate themselves, addressing the social and structural drivers of the HIV epidemic in line with their main responsibility and function.
With exception of one, the non-health sector departments included in our study developed HIV plans highlighting how HIV and AIDS impacts on the work of the department, and also reflecting on how the work of the department contributes in increasing vulnerability to HIV. Each plan details HIV programmes and activities that will be implemented in the departments. While we note the progress made in mainstreaming HIV in government departments, we also argue that the NSP call "to all government departments to mainstream HIV" is vague in that it assumes that all departments would simply define and locate their role in the response to HIV. Our data has shown that, that is only possible in some but not all departments. An example of the department of human settlement in our study is one of many others (departments) who are unclear of their role beyond a workplace response to HIV; which limits the effectiveness of the mainstreaming approach. Currently, the NSP does not provide the much-needed guidance on the 'how' of HIV mainstreaming in practice. A non-prescriptive, but guiding framework can assist in clarifying the comparative advantage of departments, and ensuring that everyone understands their role and responsibility. This we argue is an important lesson that needs to be taken forward as we think about implementing the sustainable development agenda of "leaving no one behind".
While we commend DPSA for developing guidelines to assist government departments, the guidelines are limited in guiding the process of effective mainstreaming of HIV and development of an integrated public sector response. Measures were put in place to facilitate coordination of the departments' responses including conducting consultative processes, communication and information sharing to get buy-in and to clarify roles and responsibilities. However, even with detailed planning, directorates within departments continue to work in silos, and poor reporting systems that are not synchronized exist at national level. No deliberate plans or defined steps are in place to ensure that the mainstreaming approach augments multisectoral action within the public sector in South Africa. Formulation of plans and design of HIV programmes are necessary but not sufficient conditions for effective mainstreaming of HIV (Elsey, Tolhurst, & Theobald, 2005;Rugalema & Khanye, 2004). Mainstreaming of HIV is currently undertaken as a compliance issue. The only indicator used to measure progress on the target of mainstreaming HIV in government departments in the NSP is "number of departments that have developed and have their plans approved" and nothing said about coordination, tracking implementation and outcome of the department's HIV plans (Republic of South Africa, 2012). The literature notes that effective HIV mainstreaming constitutes an evolving process and not an end in itself. HIV mainstreaming has to be undertaken as a continuous process with feedback and policy adjustment, and monitoring and evaluation. It is not just a tick-box exercise to assess whether a plan is there or not (Kenyon, Heywood, & Conway, 2001;Rugalema & Khanye, 2004). Challenges of vertical coordination of the HIV response in South Africa were also noted in other studies where a silo approach was observed in the HIV monitoring and evaluation system, promoting divisions between spheres of government (Kawonga, Blaauw, & Fonn, 2012).
Data from our research also highlighted poor collaboration in the response to HIV between government departments. The currently used approach to mainstreaming fails to promote partnerships and joined-up government in the response to HIV within the public sector. The DPSA guidelines are useful in informing the process of developing individualised departmental HIV plans, but limited in regard to partnerships and collaborations on HIV programmes within the public sector. Coordination of action within the public sector is a critical component of a multisectoral response to HIV and AIDS. It would allow for joined-up government, where departments would interconnect, complement one another, pool resources, and develop an effective HIV response (Ling, 2002;Pollitt, 2003).
Other than ISHP of DBE, DSD and DoH, we found no other evidence of partnerships and collaborations in HIV programmes between government departments included in our study. We noted in the review of plans that some programmes were duplicated across departments. For example, more than one department had a similar programme for orphans and vulnerable groups, yet no collaboration which would have allowed for efficient use of resources. Respondents in KIIs alluded to the difficulty of measuring impact of programmes when multiple actors are involved, which they understood to be one of the factors that explains poor collaboration in HIV programmes. We also found that lack of participation by some government departments in platforms that are meant to create space for collaboration between departments, renders these forums ineffective, as was observed elsewhere (Mahlangu, Vearey, Thomas, & Goudge, 2017).
HIV mainstreaming in silos (within departments), without collaboration (between departments) is limited in its ability to contribute towards an effective multisectoral response to HIV, and even in its ability to contribute towards meeting the 2030 sustainable development agenda of ending AIDS. Furthermore, the success of treatment as prevention, and achieving the 90-90-90 target is highly dependent upon Step 2 Vertical coordination between national level Step 2 Vertical coordination between national level Step 2 Vertical coordination between national level Step 2 Vertical coordination between national level with provincial and local level with provincial and local level with provincial and local level with provincial and local level Step 1: HIV Mainstreaming Each sector considers what positive or negative impacts the sector operation may have on the spread and consequences of HIV/AIDS; the specific risks HIV poses for its work; what activities and interventions can be implemented to mitigate risks, protect workers and communities; and what contributions can be made to the national response.
Step 3: Horizontal coordination between sector departments (joined-up government response to HIV)

Figure 2:
Steps towards effective mainstreaming of HIV in public sector departments people adhering to their treatment, and we know that adherence is determined by several social and structural factors that cannot be addressed by one sector alone (Moshabela et al., 2016;Orne-Gliemann et al., 2015). Multisectoral action remains crucial in scaling up an effective response to HIV and AIDS (Gavian et al., 2006;Jerling, Pelletier, Franzo, & Covic, 2016). Based on the findings of this study, we propose a three-step approach to an effective HIV mainstreaming approach (Figure 2).
The steps include not only development of mainstreamed HIV plans, but ensuring that the plans are coordinated between levels of government, and that collaborations and partnerships on HIV programmes are established to augment multisectoral action on the HIV response in the public sector. This will require strengthening mechanisms and platforms for coordination and collaboration for an integrated public sector response to HIV. Such a response promotes a partnership that draws on the comparative advantage of departments involved. Interventions that would enable such a partnership need to be urgently defined for a public sector response to HIV in South Africa to make a meaningful contribution towards achievement of the target of ending AIDS by 2030. The focus should not only be about getting sectors involved, but also about ensuring that efforts are coordinated within and between departments. Only then will the mainstreaming approach augment multisectoral action that will facilitate realisation of the SDG agenda of "leaving no one behind" and achievement of the national and global targets on HIV that are embedded in collaborative efforts. Other strategies that were suggested during KIIs included strengthening of integrated planning, budgeting and reporting processes between levels of government (local, provincial and national government). Clarification of roles and responsibilities of different departments was also highlighted as key to avoid overlaps and confusion amongst actors involved. Improved participation by government department in intergovernmental structures such as the AIDS councils was also recommended for better communication and coordination of efforts. There was also a suggestion about the need to shift the culture of operating in isolation amongst departments. Emphasising the need to get departments to understand the notion of a whole of government approach, where it will not be about a single department claiming credit, but all departments contributing towards one whole, "a public sector response to HIV".

Conclusion
Our study has shown that the public sector is hugely involved in the response to HIV and AIDS in South Africa. Progress has been made in mainstreaming HIV and AIDS through development of HIV plans or strategies in non-health sectors. However, this paper cautions against the currently adopted approach that unintentionally promotes working in silos and does not encourage pooling and effective utilisation of resources within the public sector. Such an approach is limited in its ability to augment the multisectoral response to HIV, and in its contribution towards realising the SDG agenda and the global targets on HIV. We proposed a three-step process towards effective mainstreaming of HIV. The proposed process is not only focused on drawing in non-health sector departments in the response to HIV, but also on ensuring that there is coordination and collaboration on HIV and AIDS programmes within and between departments. Such an integrated public sector response will make a meaningful contribution to the national response, and ultimately contribute towards meeting the SDGs and the global targets on HIV and AIDS. A framework for collaborative action and joined-up government on the response to HIV is urgently needed.