The role of policy actors and contextual factors in policy agenda setting and formulation: maternal fee exemption policies in Ghana over four and a half decades

Background Development of health policy is a complex process that does not necessarily follow a particular format and a predictable trajectory. Therefore, agenda setting and selecting of alternatives are critical processes of policy development and can give insights into how and why policies are made. Understanding why some policy issues remain and are maintained whiles others drop off the agenda is an important enquiry. This paper aims to advance understanding of health policy agenda setting and formulation in Ghana, a lower middle-income country, by exploring how and why the maternal (antenatal, delivery and postnatal) fee exemption policy agenda in the health sector has been maintained over the four and half decades since a ‘free antenatal care in government facilities’ policy was first introduced in October 1963. Methods A mix of historical and contemporary qualitative case studies of nine policy agenda setting and formulation processes was used. Data collection methods involved reviews of archival materials, contemporary records, media content, in-depth interviews, and participant observation. Data was analysed drawing on a combination of policy analysis theories and frameworks. Results Contextual factors, acting in an interrelating manner, shaped how policy actors acted in a timely manner and closely linked policy content to the intended agenda. Contextual factors that served as bases for the policymaking process were: political ideology, economic crisis, data about health outcomes, historical events, social unrest, change in government, election year, austerity measures, and international agendas. Nkrumah’s socialist ideology first set the agenda for free antenatal service in 1963. This policy trajectory taken in 1963 was not reversed by subsequent policy actors because contextual factors and policy actors created a network of influence to maintain this issue on the agenda. Politicians over the years participated in the process to direct and approve the agenda. Donors increasingly gained agenda access within the Ghanaian health sector as they used financial support as leverage. Conclusion Influencers of policy agenda setting must recognise that the process is complex and intertwined with a mix of political, evidence-based, finance-based, path-dependent, and donor-driven processes. Therefore, influencers need to pay attention to context and policy actors in any strategy.


Background
The development path of health policy, whether as intent, a practice, or a written document, can be difficult to predict because it is a complex and intertwined process and does not necessarily follow a particular format. Understanding why some policy issues remain and are maintained while others drop off the agenda (agenda setting and selection of alternatives) is an important field of enquiry since it can give insights into this complex process. This is because getting and maintaining policy issues on the agenda is an essential part of decisions made during policy development.
Green-Pedersen and Wilkerson [1] argue that the explanations proposed for why some issues make it onto the agenda and others fail are wide ranging. Some are structural, emphasizing how institutions are organized to advantage some alternatives or issues over others. Some are cognitive, emphasizing how individuals or even institutions process information in ways that limit what will be addressed at any given time. Others emphasize the role of external events or public opinions, and how they can combine with political incentives to quickly shift attention to a new direction [1].
Some issues, once on the agenda, are maintained over time and periodically re-examined to maintain their recurrence [2]. Political attention of vote-seeking politicians, for example, maintained health policy issues on the national agenda over time in Denmark and the United States [1]. There is, however, very little research related to how and why some policies have a long life and are maintained over time despite periodic threats to their existence; while others cease to exist.
The aim of this paper is to advance understanding of health policy agenda setting and formulation in low-and middle-income country (LMIC) settings by exploring how and why maternal (antenatal, delivery and postnatal) fee exemption policy agendas in the health sector in Ghana have been maintained over the four and half decades since a 'free antenatal care in government facilities' policy was first introduced in October 1963. Specifically, we ask: How have maternal user fee exemption policies evolved in Ghana since independence? Which actors have been involved in the policy agenda setting and formulation and why? What contextual factors influenced the process over time, how and why?
Advancing the understanding of policy agenda setting and formulation process, especially how and why a policy agenda item is maintained over time, is an essential area of analysis to inform public social policy development and implementation. Nevertheless, there is limited research and publications on policy analysis in LMICs [3] and in particular on processes of agenda setting and formulation [4]. Our work firstly contributes to the general understanding of policy agenda setting and formulation processes in a LMIC setting. Secondly, it provides insights on how and why maternal fee exemption policies in Ghana were maintained over four and half decades despite the existence of at least eight distinct threats or opportunities for major policy reforms.

Ghana health sector
The Ghanaian health sector has had a hierarchical, predominantly publically financed, publically administered and delivered, services model since independence in 1957. However, a strong private sector participation in service delivery has always accompanied it. Out-of-pocket payments at point of service have also ensured continuing 'private' financing. The sector underwent two major reforms in the 1990s. These were the creation of the Ghana Health Service (GHS) under the Ghana Health Service and Teaching Hospitals Act; and the adoption of a sectorwide approach in 1997. Prior to passage of the Ghana Health Service and Teaching Hospitals Act in 1995, the Ministry of Health (MOH) was the regulator of public and private sector, the body responsible for health sector policy direction, coordination, monitoring and evaluation, and the provider of public sector services. The Ghana Health Service and Teaching Hospitals Act 525 created an agency model in the health sector. The MOH became a civil service ministry responsible for overall sector policy making, coordination, monitoring, and evaluation, with the GHS providing public health and clinical services [5,6]. Under the sector-wide approach, dialogue between government and international donors shifted up a level: from the planning and management of projects, to the overall policy, institutional, and financial framework within which health care is provided at national level [7]. The Government of Ghana, represented by the MOH, and international donors jointly agreed to national priorities expressed in the programme of workwhich states the policies, strategies, targets, and resource envelope and allocation for the sector [8,9].
In the immediate post-colonial period (March 1957) and several years afterwards, the majority of policy agenda and formulation decisions were undertaken mainly by politicians and a small group of bureaucrats [10]. The sectorwide approach created a new avenue for policymaking platforms between the MOH, international donors, and other actors broadening the scope and range of policy actors. As a result, expertise could be drawn from other actors in or outside the health sector to form groupings to guide the process. Yet, the ultimate policy choice still rested with politicians and a few bureaucrats [11]. A handful of policy elites taking the ultimate decision is not peculiar to Ghana. In their work on developing countries, Grindle and Thomas [12] noted that small policy elitesgovernment officials and civil servantsstrongly influenced the agenda and the nature of adopted policies.

Methods
A longitudinal mix of historical and contemporary case studies of policy agenda setting and formulation for a specific issuefee exemptions for maternal health serviceswas conducted for the period 1957 to 2008. The case study approach was ideal since it allowed collection and analysis of comprehensive, systematic, and in-depth information within a real life context [13,14]. Nine specific fee exemption policy agendas for maternal health have been set since independence in 1957 and each of these was treated as a separate unit of analysis or case.
To systematically attempt to reconstruct the dynamics surrounding the nine historical maternal fee exemption policy agenda setting and formulation events, we relied on mixed methods, and analysed data in the light of an appropriate conceptual framework. Data was collected between June 2012 and May 2014 using key informant in-depth interviews, a desk review of documents and archival materials including media content from independence (1957) through to 2008, and participant observation during a 20 month period of practical attachment at the Policy Planning Monitoring and Evaluation (PPME) directorate of the MOH by one of the authors (AK). The PPME is responsible for the coordination of policy formulation and strategic planning for the health sector. Participant observation there was therefore ideal for observing and understanding aspects of the processes involved in contemporary policy agenda setting and formulation.
The focus of the in-depth interviews was to obtain real-life experiences of policy agenda setting and formulation processes from respondents. In total, 27 national level respondents were interviewed based on a semistructured interview guide. Fifteen of these respondents were identified from health sector documents reviewed, while the rest (12) were suggested by other respondents. The in-depth interviews were conducted via face-to-face meetings, e-mails, and phone. Respondents included actors within government settings such as past and current officials of the MOH (10), the GHS headquarters (3), the National Health Insurance Authority (4), and a former Minister of Health (1). Respondents also included actors outside government settings such as officials of the Christian Health Association of Ghana (1), the Coalition of Non-Governmental Organizations in Health (1), international donors (4), and health professional bodies (3). Interviews were tape-recorded and later transcribed verbatim by a neutral person to maintain the original messages of respondents. Where permission was not granted to tape-record an interview, notes were taken and verified later with the respondent. All transcriptions were read and analysed repeatedly and organized into retrievable sections based on the analytical framework. Document and archival review and analysis were used to map the historical sequence of events, identify policy actors, and further triangulate findings with respondent's information. The study greatly relied on varied documents to trace historical happenings. Documents were assessed based on four criteria developed by Scott [15]. Firstly, authenticity, which assesses that the evidence is genuine and of unquestionable origin. Secondly, credibility, which assesses whether the evidence is free from error and distortion. Thirdly, representativeness, which assesses whether the evidence is typical of its kind, and, if not, whether the extent of its untypicality is known. Finally, meaning, which assesses whether the evidence is clear and comprehensible [15]. National archives, the National Parliament Library, the George Padmore Research Library, and the Ghana Publishing Corporation were the sources of data for health legislative documents such as National Decrees, Acts of Parliaments, and National Regulations; old health-related reports and records of one national newspaperthe Daily Graphic were also used. We obtained access through the policy analysis unit of the MOH to archives of nonconfidential official documents including letters, meeting minutes, memoranda, health sector review reports, health sector programme of work, national strategic plans, and agreements related to decisions to provide maternal user fee exemptions. Additionally, the web-based search engine Google Scholar was used to obtain published literature related to maternal fee exemptions. Relevant sections of all reviewed documents were highlighted and coded based on the categories identified in the analytical framework.

Analytical framework
To guide the analysis of the data we drew on several policy analysis theories, frameworks, and concepts in the literature. Grindle and Thomas [12] conceptualize context as including the structure of class and interest group mobilization in the society, historical experiences and conditions, international economic and political relationships, domestic economic conditions, the administrative capacity of the state, and the impact of prior or conterminously pursued policies. They also include in context, the individual characteristics of policy actors such as their ideological predispositions, professional expertise and training, memories of similar policy situations, position and power resources, political and institutional commitments, loyalties, and personal attributes and goals. They observe that policy actors are never fully autonomous. Instead, they work within several interlocking contexts that confront them with issues and problems they need to address, set limits on what solutions are considered, determine what options are feasible politically, economically and administratively, and respond to efforts to alter existing policies and institutional practices.
Kingdon's [16] theory and framework of agenda setting argues that active participants (policy actors) and the processes by which agenda items and alternatives come into prominence are key factors that affect policy agenda setting and choice. Policy actors in his USA study included the President, the Congress, bureaucrats in the executive branch, and various forces outside of government including the media, interest groups, political parties, and the general public. Policy agenda setting and choice processes are embedded within their context and, as such, influence how policy actors operate within these processes.
Power is a key factor in health policy processes [17]. Contextual factors may serve as a source of power to influence policy actors' action, inaction, and choice. Policy actors therefore can become influencers within a specific context to affect policy agenda setting and formulation processes. As noted by Mintzberg [18], to be an influencer, one requires some source of powerdefined by control of a resource, a technical skill and body of knowledge, or stemming from a legal prerogativesor authority, coupled with active involvement in ongoing processes in a politically skilful way.
Drawing on these concepts of context, policy actors, and power, we attempted to systematically reconstruct nine historical agenda setting and policy formulation events. Working iteratively on data gathered, patterns, themes and categories that emerged were tabulated and further analysed. The analysis process involved mapping to our analytical frameworkcontextual situations, policy actors and their role, linkage among policies, specific policy content, power sources and how these influenced the agenda setting processes and why. We acknowledge the problems involved in mapping the exact sequence of events. To minimise this, varied sources of data were used to reconstruct, insofar as possible, the chronology and dynamics of maternal fee exemption policies agenda setting and formulation processes.

Ethical considerations
This study forms part of a larger study -' Accelerating progress towards attainment of Millennium Development Goals 4 and 5 in Ghana through basic health systems function strengthening'for which ethical approval was granted by the GHS Ethical Review Committee and the School of Social Science Research Assessment Committee of Wageningen University and Research Centre. Informed consent was obtained from all respondents, and respondent's anonymity was maintained and protected using codes as labels during the study.

Results
This section contains a historical reconstruction of the dynamics related to the nine maternal fee exemption policy agenda setting and formulation events insofar as possible. We acknowledge the difficulty in providing a full explanation of events as they unfoldedreconstructing who said what, when, to whom and how it was received. Where such data is available, it is duly noted; otherwise, the gap is noted and possible inferences are made from interpretation of data.

Policy actors and agenda setting
Maternal fee exemption policies studied included free healthcare services related to one or more of antenatal, delivery, and postnatal services, starting from the initial introduction of free antenatal service in 1963. Policies related to maternal fee exemption were maintained and modifiedincluding expansions and contractions; but were never completely dropped over the period studied. Nine specific maternal fee exemption policies were identified along the pathway, as the policies evolved from user fee exemption to national health insurance premium exemption. Table 1 summarises the maternal fee exemption policies historical timelines, policy instruments and policy contents between 1963 and 2008.
Over the period studied, we classified actors involved in maternal fee exemption policies based on their primary role into four groups. The first group, 'policy agenda directors' , includes high level politicians such as heads of state who gave directives to either set the maternal fee exemption agenda or modify a previously existing policy. The second group 'policy agenda approvers' includes high and middle level politicians such as heads of state and ministers of health who gave approval for existing maternal fee exemption policies to be maintained and/or modified. The third group, 'policy agenda advisers' , includes government and non-government individuals and organizations who advised agenda directors and approvers. Policy agenda advisers includes the Ministry of Health and its agencies such as the GHS and National Health Insurance Authority (NHIA), as well as those outside the health sector such as the Attorney General Office and National Development Planning Commission. Non-government policy agenda advisers include international bilateral and multilateral donors. Policy agenda advisers provided technical expertise in varying capacities to push/keep particular ideas on or off the agenda. Some have, over the period studied, provided financial resources to support their ideas and in some cases, set the agenda. The fourth group, 'policy agenda advocates' , includes those who have supported and campaigned directly or indirectly to maintain maternal fee exemption policies. Examples include the general public, the Ghana Medical Association, and the Pharmaceutical Society of Ghana.

Contextual factors and agenda setting
Context and policy actors consistently influenced the manner in which policy agenda setting and formulation related to maternal fee exemptions occurred over the period of study (Table 2). Contextual factors that shaped maternal fee exemption polices from 1963 to 2008 included political ideology, economic crises, historical events, change in government, election years, austerity measures, international agendas, and country-based health outcomes in the form of health demographic indicators. These contextual factors also served as sources of power that policy actors used to influence the agenda setting and formulation processes, and justify their actions and inactions. They are described below for each of the nine discrete policy change periods we identified.

Free Antenatal Care in the Public Sector Directive
Prior to independence in March 1957, patients paid charges for hospital services. The existing health law -Hospital Fees Ordinance, Regulation Number 56 of 1942stipulated schedules of fees for hospital services [19]. In the context of political emancipation and the euphoria that marked independence, it was evident that charging of fees for services was at odds with the political ideology of free health and educationthe Nkrumahism social philosophy [20] -promoted by the first head of state, Dr Kwame Nkrumah [21]. Thus, the first financing policy related to maternal health services, the 21st October 1963 directive by the MOH that with immediate effect, all antenatal services should be provided at government hospitals free of charge [19], had as its main contextual agenda driver, ideology. Public reminders of this popular directive to provide free health services for all were carried in national newspapers with headlines such as; 'hospital fees, no charge'; 'free health service' and 'free medical service soon' [21][22][23]. However, the MOH used a piecemeal approach in making free health for all a reality, although it was a political 'Except in respect of accommodation and maintenance fees specified in the Second Schedule to this Decree and subject to any other provision of this Decree, no fees shall be paid in a hospital by -(b) any persons in respect of antenatal care at a Clinic or Health Centre; -(c) any multiparous patient with a history of five or more pregnancies, or any patient referred to a maternity or other hospital from a clinic or health centre or any patient referred to any such hospital by a registered midwife or registered medical practitioner'      directive. In addition to the free antenatal service, the MOH also made adjustments to reduce existing hospital fees and provided free care for other services. In addition to blacklisting all existing policies for political reasons, an important contextual factor motivating removal of fee exemptions was the worsening economic situation. The government of Ghana was faced with declining economic indicators and increasing health expenditure. By the mid-1960s, the economy was stagnant. Gross national income per capita was US$ 200 in 1963 and only slightly increased to US$ 220 in 1969 [24,25], resulting in a health sector budget deficit in the face of increasing expenditure due to hospital fee exemptions, minimal fees for other health services, and an increasing population. Despite these motivations to reintroduce hospital fees, free antenatal service was captured as a user fee exemption policy within the Hospital Fees Decree. Additionally, delivery service for multiparous patients and patients referred to a hospital or clinic by a registered midwife or medical practitioner was made free. How and why did maternal fee exemptions remain on the agenda?
In designing the hospital fees policy content, the MOH bureaucrats collated proposed reasonable fees from all government hospitals. These proposed fees were agreed on in a consultative meeting with regional heads of government health services and submitted to the NLC military government for approval [26]. Pending approval by the NLC, the MOH sent a circular, dated 6th February 1968, to all government health facilities in an attempt to regularise the varying charges that the government's decision to reintroduce hospital fees had caused. The implementation of these new charges brought an uproar from the general public [19]. Patients had to pay both dispensary fees and the cost of medicines and some facilities were charging more than stipulated [19,27]. The social uproar caused the MOH to issue a press release on 6th July 1968 to suspend the operation of the proposed charges [19]. The suspension only delayed rather than altered the government's intention to reintroduce hospital fees. Subsequently, the approved fees were introduced with the Hospital Fees Decree of 18th June 1969, to be implemented from the 1st October 1969 [19,27,28].
During this period, however, MOH bureaucrats as policy agenda advisers, advocated for maternal fee exemption in the Hospital Fees Decree, and this was approved by the NLC government. According to the MOH, this was done to pacify the general public and minimise the financial burden of care. Another critical contextual factor that influenced the MOH decision was evidence from the health management information system about the high number of maternal deaths. A former MOH staff stated: 'In 1969, we [MOH] realised that maternal mortality was high and that we had to do something about it…' [Interview, 15/7/ 2013]. Furthermore, the NLC military governmentpolicy agenda approverdid not fully ignore the social unrest against the reintroduction of government hospital fees and the need to consolidate political power and gain acceptance among the general public in approving the maternal fee exemptions.
The NLC military government handed over to a democratically elected government led by Prime Minister Dr KA Busia on 1st October 1969, the same day the implementation of the Hospital Fees Decree was to start. Although, the Decree was softened to pacify the agitated public, its implementation was still vehemently opposed causing another social unrest. As a result, the Busia-led administration suspended implementation of the Decree and set up a five-member committee, known as the Konotey-Ahulu committee, comprising a medicine and therapeutics lecturer, an industrialist, a health worker unionist, an Arts Council national organiser, and a general medical practitioner. The committee was tasked to investigate hospital fees and recommend appropriate charges for health care services in government facilities [19].
With the suspended Hospital Fees Decree, Ghana was back to charging minimal hospital fees much lower than the actual cost of service delivery; although, it was experiencing increasing health care expenditure and a declining economy [29]. Thus, in 1970, the MOH advised Busia's government: 'free health service is not the way to go, with increasing health bill and reduced revenue' [Former MOH staff, 15/7/2013].
As in the previous reform, an important contextual driver was empirical evidence on the performance of the economy. Following the MOH's advice, Busia's government decided to reintroduce hospital fees [30]. The new Hospital Fees Act 387 was passed into law by the National Assembly. The Act 387 reflected the repealed Hospital Fees Decree. Existing fee exemptions for maternal health care under the Hospital Fees Decree were maintained by the national assembly and government [31]. However, before the Busia government could develop a Legislative Instrument (LI) to interpret the Act 387 with specific fees, it was ousted on 13th January 1972 in a military coup.
'The LI was to be based on the committee's report; however, Busia was overthrown before his government could implement any of the 65 recommendations of the Konotey Ahulu committee report' [Former MOH staff, 15/07/2013].
The Military Government that replaced the Busia government was known as the National Redemption Council (NRC). The NRC was in a dilemma as to whether to charge hospital fees or abolish them. As a result, the NRC commissioner for health invited views from the public on the recommendations of the Konotey-Ahulu Committee [32]. The Konotey-Ahulu Committee in 1970 had recommended that there could be no health service without fees. For maternal services, it recommended that antenatal care should no longer be free, and that fees be paid towards the cost of medicines dispensed in government health facilities for maternal services. Also, multiparous patients with a history of five or more pregnancies should bear some cost for their health care services; not everybody agreed with this view. For example, a Daily Graphic newspaper correspondent was of the opinion that charging fees would scare away people and an ignorant expectant mother would totally refuse to attend hospital knowing that she would be charged [33].
There were over 4 months of public debate on whether to charge hospital fees or not [33][34][35][36]. A review of Daily Graphic newspapers from 1973 to 1974 revealed that the majority of correspondents recommended the government to charge hospital fees and exempt the poor and unemployed. Despite these recommendations, the NRC government did not implement the Hospital Fees Act 387 and the Konotey-Ahulu Committee's recommendations. Charging hospital user fee was still unpopular with the general public, although their disapproving voices were not expressed greatly by the Daily Graphic correspondents. The NRC military government, in order to consolidate political power and gain acceptance, did not implement the Hospital Fees Act 387 and the Konotey-Ahulu Committee recommendations. The frequent change in government during this turbulent period was accompanied by the country moving from economic decline to disaster as gross domestic product per capita fell from US$ 281 in 1970 to US$ 180 in 1983. State institutions and public services were gravely damaged and under-resourced [37]. The decline in the health budget led to a reduced capacity to procure medicines and consumables. By the early 1980s, deteriorating health care services deterred the general public from using government facilities and some patients only used these facilities when their health conditions were critical. Medical and pharmaceutical professional bodies advocated for the introduction of hospital fees to revive falling standards of health care and threatened to strike [38][39][40]. Health workers unilaterally introduced de facto hospital fees as a result of the economic crisis and declining availability of health service inputs [40]. All these contextual factors combined to put hospital fees back on the agenda by the early 1980s. However, once again, fee exemption for maternal health service was maintained on the agenda. How did maternal user fee exemption policy survive the urgent need to reintroduce hospital charges for all services in the face of economic crisis?
To regularise the fees already charged by government health facilities, the MOH conducted a study to propose hospital fees to the PNDC military government. The PNDC government was initially not fully supportive of hospital fees. They reduced the amounts proposed, and later approved them. The reason for the initial reduction was political: 'The PNDC representative for finance said it will be ill politics to introduce user fees when the country had the economic crunch at the time. The proposed fees were reduced by about 90 %' [Former MOH staff, 22/8/2012].
The reason for the later approval was the further decline in health budget as a result of economic crisis.
To legitimise the approved fees, the MOH drafted the Hospital Fees Regulation with the assistance of the legal department of the Attorney General's office. The initial draft made no exemption for maternal health service. This was contested by the United Nations International Children's Emergency Fund (UNICEF): 'The first time we introduced the regulations, UNICEF was against our fees because for them the policy is that maternal care should be free. We argued that free maternal services would defeat family planning purposes' [Former MOH staff, 22/8/2012].
The MOH, therefore later incorporated fee exemptions for antenatal care, postnatal care and treatment at child health welfare into the new Hospital Fees Regulation, in part, because UNICEF advised against maternal hospital fee charges and demonstrated further interest in free maternal care by providing financial support to procure folic acid for antenatal care. Also, international maternal and child health discourse influenced the decision, in the sense that, in the early 1980s, maternal and child health attention had shifted with more focus on child health and family planning [41]. Some family planning activities were incorporated into postnatal care services [42,43], making free postnatal care a viable policy. The resulting Hospital Fees Regulation (LI 1277), came into force on 21st April 1983, approved by the PNDC military government [44].
Up until this point, development partners (donors) had not played visible and significant roles in shaping policy agendas related to user fee exemptions in Ghana. The appearance on the scene of UNICEF in a strong role as an agenda influencer was a reflection of the increasing amounts of development partner project aid flowing into Ghana as into many other LMICs because of the economic crisis and international development policies of the seventies and eighties.

Hospital Fees Regulation (LI 1313)
Evidence of Ghana's continuing economic decline drove, in part, the next agenda. Economic decline still posed a major challenge to Ghana and the PNDC government turned to the International Monetary Fund (IMF) and World Bank. By April 1983, government had introduced a Structural Adjustment Programme under the auspices of the IMF and the World Bank. This economic recovery policy implemented over 3 years, from 1983 to 1986, was intended to halt the downward economic spiral and stabilize the economy on a reasonable track [29,45].
Hospital fees were substantially increased in July 1985, partly on the recommendation of the IMF and World Bank under the Structural Adjustment Programme [40] and partly because the existing fees could not recover costs and health facilities had already increased their fees to halt further decline of health care services [46,47]. Although, hospital fees were increased with the aim of full cost recovery, antenatal and postnatal user fee exemptions were maintained and mentioned in the Hospital Fees Regulation (LI 1313). A key informant explained that with time, user fee exemption policies became a safety net for the poor and so the MOH maintained these.
'Based on experiences within the health service, fee exemptions had become a safe net for the poor so we (MOH) maintained it' [MOH staff, 27/9/2012].
1997 Presidential Directive to expand free antenatal and postnatal care to include deliveries In 1992, the PNDC allowed multiparty democratic election to be held. The PNDC re-organized itself into a political party, the National Democratic Congress (NDC) and won the December 1992 election as well as the December 1996 multiparty election 4 years later with Flight Lieutenant Rawlings as its flag bearer. In January 1997, at the beginning of his second term, the President gave a directive to include delivery service in the existing maternal (antenatal and postnatal) user fee exemption policy [48].
The President acted to mitigate the social consequences of the structural adjustment programme as evident by decreasing utilization of maternal health services and worsening health outcomes. Ghana Demographic Health Survey 1993 empirical evidence revealed that national utilisation of free antenatal service was high at 86 %. The picture was, however, different for supervised delivery, as national level supervised delivery in health facilities was only 44 %. About half of the women who received free antenatal care did not return to deliver in those health facilities [49] partly because of their inability to pay at the point of use. Additionally, the MOH 5-year programme of work, attributed a high national maternal mortality rate estimate of 214 per 100,000 a live births to the harsh economic recovery policy of the structural adjustment programme [50]. These were major drivers of the agenda to provide free maternal (antenatal, delivery and postnatal) services. Nevertheless, the maternal user fee exemption guideline developed by the MOH to implement the directive in November of the same year provided fee exemption for only four antenatal visits; further visits had to be paid for as well as deliveries and postnatal care. Why did decision makers water down the intent and miss this opportunity to more radically reform existing maternal user fee exemptions?
Interviews with a key informant explained that policy agenda advisers and formulators in the MOH limited the scope of the policy based on their experience, analysis and judgement of what was contextually feasible and practical to implement at that time.
'Based on our [MOH] experience of reimbursing bills for exemptions, the annual budget allocated was not sufficient to foot the bill; as such we could not have added delivery services' [MOH staff, 27/9/2012].
There was insufficient government financial support to fully implement the directive [51]. This is because by the mid to late 1990s, Ghana's structural adjustment programme efforts had faded with slowed economic growth [24] contributing to a reduction in the allocation of government budget to the health sector and hence the subsequent inability to fully implement the directive.

Maternal delivery exemptions in four selected regions
The NDC lost the December 2000 election to the New Patriotic Party led by Mr John Kufuor, the new face of the Danquah-Busia tradition, which was the party in opposition at independence in 1957 and had briefly ruled the country from 1970 to 1972 before it was ousted in a military coup [11]. The Kufuor government came to power in a context of stagnant economic and even regressive growth [24,52]. For example, gross national income per capita for 2002 was US$ 270; the same as in 1971 [25].
To address economic stagnation, in 2001, the government opted for debt relief under the Heavily Indebted Poor Countries (HIPC) initiative on the advice of the World Bank and IMF. This initiative was launched by the World Bank and IMF in 1996 [52]. One of the HIPC austerity measure conditionalities was for Ghana to develop a comprehensive poverty reduction strategy directed towards attainment of anti-poverty objectives consistent with the Millennium Development Goals (MDGs). As a result, relatively poorer regions (Northern, Upper West, Upper East and Central) were set to benefit most from the initiative [53,54]. As per the poverty reduction strategy, health related targets to reduce maternal mortality and under five mortality proposed by the policy agenda advisersthe National Development and Planning Commissionwith the assistance of the World Bank and IMF, favoured these regions. In this regard, the existing maternal fee exemption policy was extended to include delivery and postnatal services and geographically limited to the four deprived regions [55]. In 2004, 27 billion cedis (US$ 3.1 million) from the HIPC grant were budgeted for this purpose [56].

Expansion of maternal delivery exemptions to the whole country
The December 2004 presidential election presented an opportunity for policy actors to modify existing policies putting maternal fee exemption back on the agenda. At the December 2004 health summit meetings, the MOH and stakeholders argued that a national maternal mortality rate of 503 per 100,000 b live births was high, and there were pockets of extreme poverty across the country and not only in the regions labelled as deprived. A national user fee exemption for antenatal, delivery and postnatal services was therefore proposed by the MOH and stakeholders to help reduce maternal mortality [57,58]. Politically, this idea was approved by the government and, in 2005, 30 billion cedis (3.4 million US$) from the HIPC grant was budgeted and allocated to implement the policy nationwide [59]. Empirical evidence of high maternal mortality was thus a major agenda driver.

Integration of maternal fee exemptions into the National Health Insurance Scheme
The New Patriotic Party government won a second term in the December 2004 election. By this time, implementation of their popular promise to replace the health sector 'cash and carry' system with a national health insurance scheme and assure access to basic clinical service for all Ghanaian regardless of ability to pay had started [11].
After 2005, maternal fee exemption policy implementation suffered a major setback. Empirical evidence from evaluation of the maternal user fee exemption policy in 2006 revealed that the policy contributed to a major increase in supervised deliveries, but was significantly underfunded [60]. Issues of inadequate funds, sustainability and inability to predict when reimbursement would be paid by government were well-known and discussed within the health sector [61][62][63][64]. Not only was there a problem of inadequate funds, but also health facilities exemption bills over time exceeded the budget allocated to implement the policy. By 2007, health facilities had to stop providing free maternal health services as unpaid reimbursement bills piled. As a key informant stated: 'Maternal fee exemption became unsustainable because every month the bill was going up and going up, and it got to a point, the facilities were bringing the bill and we  [65]. To this end, the MOH, at the April 2008 Health Summit, declared maternal health a national emergency [66].
Immediate decisions and actions followed the declaration. A ministerial task force was formed to formulate a timed framework aimed at reducing maternal mortality and the MOH was tasked to estimate the impact and financial implications of subsidising the enrolment of pregnant women onto the National Health Insurance Schemes (NHIS) [66]. Additionally, the United Kingdom Department for International Development (DFID) submitted a brief to the Presidency through the MOH suggesting that all pregnant women be given functional NHIS membership cards to assure access to maternal health care and improve the performance towards attainment of MDGs 4 and 5. This, they argued, would be an effective, affordable and extremely popular policy with the Ghanaian electorate. The Minister's declaration created a charged atmosphere putting maternal health on the front burner with intense attention from all stakeholders.
The MOH drafted a memo to the Presidency on the current status of maternal health and possible interventions. The import of the MOH memo was to inform and prepare the President for his participation in the 'Busi- 'The DFID brief submitted to the Ghanaian MOH was also followed up through the British system and given to the British Prime Minister, Gordon Brown. During their meeting, the Prime Minister told President Kufuor that he had heard of the challenges of institutional maternal mortalities. My understanding is that Gordon Brown said it was a good idea to provide free maternal services' [Donor, 27/5/2014].
In London, President Kufuor announced to exempt all pregnant women from paying for maternal health services.
However, based on historical experiences and evidence of inadequate financial resources to implement previous maternal fee exemption policies [60], availability of funds was the foremost concern of the MOH. According to a former MOH senior official, no specific budget was allocated by the Ministry of Finance and Economic Planning for this policy before it was announced outside Ghana. In an interview with a senior politician for clarification, he said: 'sometimes you need to make the policy and later look for funds to implement it' [Former Minister of Health, 21/12/2012].
With no central government-allocated budget to implement the maternal fee exemption directive, the MOH relied on donor health sector budget support for financial commitment.
'DFID's contribution was the obvious choice since maternal health care was tagged as DFID-supported. Although DFID emphasised that their contribution was not for free maternal delivery but to support the whole health sector programme of work, the MOH went ahead and earmarked the funds to implement the directive' [Former MOH staff, 5/11/2012].
With secured funding from DFID through the health sector budget support and the preceding suggestion by the DFID to give pregnant women NHIA cards to assure access to maternal health care, the NHIA lobbied to implement free maternal health care arguing that it was competent in fund management and best positioned to implement the policy. Additionally, it already provided antenatal, delivery and postnatal services under its benefit package and as such, unregistered pregnant women could be issued with cards to enable them access health care without any waiting period. The MOH accepted the arguments.
'The need to incorporate it into the NHIS was realized later, because, when you make the national estimate for the number of expected pregnancies in a year and you look at the premium level, it would be cheaper to pay the premium for pregnant women than to pay for the services' [Former MOH staff, 5/11/2012].
By 27th June 2008, a guideline accompanying the directive to provide free maternal health care for all pregnant women was designed by the MOH, officials of the GHS, Ghana Registered Midwives Association, and the NHIA. The policy implemented through the NHIA started 1st July 2008.

Discussion and conclusions
Over the four and a half decades since some form of exemption from payment for maternal health services was introduced in 1963, fee exemptions for health service use by pregnant women has managed to remain on the policy agenda. However, it has remained on the agenda in a fluid process of ebbs and flows rather than in a static fixed form. Context and policy actors were the major influencers of the ebbs and flows.
Contextual factors that influenced the ebbs and flows were: political, such as Nkrumah's ideology, change in government, and election year; economic crises and austerity measures; health and demographic indicators; historical events; social unrest; and international agendas such as the MDGs. These contextual factors served as a source of power for policy actors to influence maternal fee exemption as a policy agenda item. We therefore reason with Erasmus and Gilson [17] that power is the heart of health policy process, as these case studies illustrate how policy actors used contextual factors as power leverage to justify their actions, inactions and choices.
Policy agenda setters (directors, approvers, advisers and advocates) acted within interrelated contextual factors, which sometimes worked as constraints and sometimes opened opportunities. We observed that interrelating context, whether a constraint or an opportunity, is used by specific policy agenda setters to influence the timeous manner in which policy content is made and how closely it is linked to the intended agenda. Our observations are in keeping with similar observations by Grindle and Thomas [12], that contextual factors working in interrelating manner can serve as a constraint and an opportunity within which policy actors manoeuvre to accomplish their goals.
Contextual factors working in an interrelating manner as a constraint, present policy agenda setters with conflicting options shaping the policy content to be made in a less timely manner and less closely linked to the intended agenda. For instance, within the context of high maternal mortality, economic decline, limited government budget allocation, and political authority and will, MOH bureaucrats had to assess options to make practical and feasible choices. Evidence of a health sector budget deficit at times overtook government's intent. This was the case in 1997; there were worsening indicators for supervised delivery as about half of the women who hitherto attended government health facilities for free antenatal services did not return to deliver in those facilities. To solve this issue, in January 1997, the President within his constitutional power, gave a political directive to provide free healthcare for pregnant women. The directive presented an opportunity to reform existing free antenatal and postnatal policy. However, the government did not allocate adequate resources due to economic decline. To this end, the policy content developed by the MOH bureaucrats in November the same year only partially reflected the intended agenda.
Contextual factors working in an interrelating manner as an opportunity, present policy agenda setters with complementary options in shaping the policy content to be made in a more timely manner and more closely linked to the intended agenda. For instance, within the context of economic decline, inequitable poverty indicators, high maternal mortality, donor financial support, election years, and international agendas, policy agenda setters defined the maternal health problem in relation to a clearly defined solution. This was the case for maternal fee exemption policies in the 2000s. In the early 2000s, the HIPC grant support proposed by international policy agenda advisersthe World Bank and IMFto mitigate the effect of economic stagnation provided an opportunity to improve poverty and maternal health outcomes in deprived regions. Policy formulators and international and national policy agenda advisers ensured that the policy content was closely linked to the intended agenda, and as a requirement to obtain the HIPC grant, the policy was made in a timely manner. Again, in May 2008, President Kufuor announced a 'free maternal health care' policy based on a proposed solution from international and national policy agenda advisers. In the light of secured funding from health sector budget support, the policy content was made and disseminated before the end of June 2008, for implementation on July 1st.
In addition to context serving as a source of power to shape policy actors actions, inactions and choices, policy actors also wield power by virtue of their political and administrative position, knowledge, experience, and financial commitment. Policy agenda setters (directors, approvers, advisers, and advocates) acted in varied influencer roles between 1963 and 2008. Politicians (policy agenda directors and approvers) and policy agenda advisers played an active role in maintaining maternal fee exemption policies over a long period. Maternal fee exemption policies therefore survived both military and civilian governments and have become sort of a national legacy. Politician's interest in maintaining the agenda may have varied; however, whether it was out of genuine concern to improve maternal health or to gain political capital and favour, their political support to decisions of maternal fee exemption was critical.
Both national and international policy agenda advisers played an active role in maintaining the policy. Though international policy agenda advisers did not have official government positions to make and implement public policies, over time, they became active agenda setters within the Ghanaian health sector. After the 1990s, international policy agenda advisersinternational multilateral and bilateral organisations and officialshave increasingly gained agenda access, and sometimes even set the agenda. Donors gained agenda access because they used financial support as leverage of what gets on the agenda and in the policy content.
Policy agenda setters, in varied ways and capacities, strived to maintain maternal health issues on the agenda. At critical moments of agenda re-set, re-examination, and modification of existing maternal fee exemption policies, policy agenda advisersacting as policy championstook decisions within boundaries of previous policy content, implementation challenges, such as inadequate funds, as well as current demands and expectations. Our finding agrees with the position of Shiffman and Smith [67], that strong champions are required to shape political priority for a particular policy initiative.
Policy agenda advisers as policy champions mobilised strategies and tactics in the form of commitment and consensus to maintain the maternal fee exemption policy on the agenda over the years. For example, the World Bank and IMF committed to reduce poverty and improve MDG-related targets, pushed for fee exemptions for maternal health services. They collaborated with the Government of Ghana and other state agencies such as the National Development and Planning Commission for a consensus on the practical details of the Ghana Poverty Reduction Strategy. Also, the MOH, over the years, built relationships with other policy actors such as donors at institutional level through interactions, consensus building, and collaboration towards policy development. These strategies are described as strategy capacity [68] and include the human and institutional capacity to build commitment and consensus toward a long-term strategy, respond to recurring challenges and opportunities, build relationships among policy actors, and undertake strategic communications with varied audiences. Strategic capacity is therefore critical for maintaining policy issues on the agenda over time.
The fee exemption policy for maternal health was maintained over the years in a path-dependent manner. The free antenatal service trajectory taken in 1963 was not reversed, despite varied policy agenda setters and contextual factors. Policy actors relied on context and on each other for financial support, expertise, experience, and political resources creating a network of influence to maintain a maternal fee exemption agenda over time. Some scholars argued that a process is path dependent if initial moves in one direction elicit further moves in that same direction [69,70]. Once maternal fee exemption was there, it was difficult to abolish because of wide popular support and later outcry over maternal mortality and international agenda such as the MDGs.
Policy agenda setters also relied on empirical evidence to inform their decisions, however, systematic reviews presumed as a 'gold standard' of evidence-based policymaking [71] were not used. Rather, country based empirical evidence from economic assessments, surveys, research reports, and health sector performance reviews were used. Policy agenda setters paid attention to this kind of evidence as one of several important contextual factors rather than the only or even the main one, and made use of this evidence to maintain maternal fee exemption policy on the agenda.
Finally, as noted by Shiffman and Smith [67], the power of actors, the power of ideas, political context, and characteristics of the issue are key for setting the global health agenda; these observations are also relevant at national level and evident in these case studies. In addition, a broader contextual environment such as financial allocation arrangements, international agenda and development partner's relationships, data about health outcomes, national administrative capacity to develop and implement policies, historical experience, and path dependency are also critical as shown in this paper.
Ghanaian health sector policy agenda setting and formulation is complex and intertwined with a mix of political, evidence-based, finance-based, path-dependent, and donordriven processes. The papers by Agyepong and Adjei [11] and Seddoh and Akor [72] note this complexity. Actors and stakeholders who want to influence agendas need to pay attention to context and policy actors in any strategy. Efforts to influence policy agenda setting must recognize that empirical evidence is only part of a complexity of factors of which context, path dependency, and politics are also very important. Moreover, the influence of evidence is dependent on awareness of its availability as well as it use for advocacy in policy agenda setting and formulation in the right window of opportunity.
As policymaking processes are relevant across other LMICs, and national policy actors are likely to confront similar scenarios, we hope this paper contributes to learning beyond Ghana in which this work was conducted to other LMIC in sub-Saharan Africa and beyond.
Endnotes a Maternal mortality rate estimated at 214 per 100,000 live births was based on Ghana Demographic and Health Survey (1993) data [73]. b Maternal mortality rate estimated at 503 per 100,000 live births was based on Ghana Demographic and Health Survey (2003) and institutional maternal mortality data [58].