76223 v3 H N P D i s c u s s i o n P a p e R MAPPING, COST, AND REACH TO THE POOR OF FAITH-INSPIRED HEALTH CARE PROVIDERS IN SUB-SAHARAN AFRICA Strengthening the Evidence for Faith-inspired Health Engagement in Africa, Volume 3 Edited by Jill Olivier and Quentin Wodon November 2012 MAPPING, COST, AND REACH TO THE POOR OF FAITH-INSPIRED HEALTH CARE PROVIDERS IN SUB-SAHARAN AFRICA Strengthening the Evidence for Faith-inspired Health Engagement in Africa, Volume 3 Edited by Jill Olivier and Quentin Wodon November 2012 Health, Nutrition and Population (HNP) Discussion Paper This series is produced by the Health, Nutrition, and Population Family (HNP) of the World Bank's Human Development Network (HDN). The papers in this series aim to provide a vehicle for publishing preliminary and unpolished results on HNP topics to encourage discussion and debate. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations or to members of its Board of Executive Directors or the countries they represent. Citation and the use of material presented in this series should take into account this provisional character. Enquiries about the series and submissions should be made directly to the Editor, Martin Lutalo (mlutalo@worldbank.org). Submissions undergo informal peer review by selected internal reviewers and have to be cleared by the Task Team Leader’s (TTL's) Sector Manager. The sponsoring department and author(s) bear full responsibility for the quality of the technical contents and presentation of material in the series. For information regarding the HNP Discussion Paper Series, please contact Martin Lutalo at mlutalo@worldbank.org or 202-522-3234 (fax). © 2012 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 All rights reserved. ii Health, Nutrition and Population (HNP) Discussion Paper MAPPING, COST, AND REACH TO THE POOR OF FAITH-INSPIRED HEALTH CARE PROVIDERS IN SUB-SAHARAN AFRICA Strengthening the Evidence for Faith-inspired Health Engagement in Africa, Volume 3 Edited by Jill Oliviera and Quentin Wodonb a School of Public Health and Family Medicine, University of Cape Town, Cape Town, South Africa b Education Department, World Bank, Washington, DC, United States Collection prepared in support of the World Bank’s work on private service delivery in human development. Abstract: As African governments, donors, and a wide range of organizations increase their efforts to reach the Millennium Development Goals (MDGs) and set the agenda for the post-MDGs era, the role of non-state providers of health care is gaining new attention. In Africa, the largest non-state networks of providers are often faith-inspired. But how important is the role of faith-inspired institutions (FIIs) in health care provision in Africa? How substantial are their market share and reach to the poor? How affordable are the services provided by FIIs to households? How satisfied are households with these services? What are some of the interesting and innovative experiences that have been documented in terms of FIIs providing quality services to underserved populations? Beyond facilities-based care, which types of non-institutionalized initiatives emerge out of communities of faith that are generative of health? How can these initiatives be mapped, understood and leveraged for better health and development? The objective of this edited series of three World Bank HNP Discussion Papers is to gather tentative answers to such questions. This third volume in the series focuses on ways to ‘map’ (in the different uses of that terminology) faith-inspired providers, and on assessment of their cost for patients and the extent to which they succeed in reaching the poor. Keywords: Health, private provision, faith, religion, poverty, Africa Disclaimer: The findings, interpretations and conclusions expressed in the paper are entirely those of the authors, and do not represent the views of the World Bank, its Executive Directors, or the countries they represent. Correspondence Details: Dr. Quentin Wodon, World Bank, 1818 H Street NW, Washington, DC, 20433. Email: qwodon@worldbank.org. iii Table of Contents ACKNOWLEDGMENTS…..………………………………………………………….…v MAPPING, COST, AND REACH TO THE POOR OF FAITH-INSPIRED HEALTH CARE PROVIDERS IN SUB-SAHARAN AFRICA: A BRIEF OVERVIEW .…….….1 Jill Olivier and Quentin Wodon CHAPTER 1 - DO FAITH-INSPIRED HEALTH CARE PROVIDERS IN AFRICA REACH THE POOR MORE THAN OTHER PROVIDERS? .………….…………..….7 Jill Olivier, Clarence Tsimpo and Quentin Wodon CHAPTER 2 - LAYERS OF EVIDENCE: DISCOURSES AND TYPOLOGIES OF FAITH-INSPIRED COMMUNITY RESPONSES TO HIV/AIDS IN AFRICA ………25 Jill Olivier and Quentin Wodon CHAPTER 3 - MAPPING RELIGIOUS COMMUNITY HEALTH ASSETS AND INITIATIVES: LESSONS FROM ZAMBIA AND LESOTHO .……………………….52 Jill Olivier, James R Cochrane and Steven de Gruchy CHAPTER 4 - MAPPING RELIGIOUS HEALTH ASSETS: ARE FAITH-INSPIRED FACILITIES LOCATED IN POOR AREAS IN GHANA? .…………………………..62 Harold Coulombe and Quentin Wodon CHAPTER 5 - COMMUNITY HEALTH ASSETS MAPPING: A MIXED METHOD APPROACH IN NAIROBI .………………………………………..…………………..76 John Blevins, Sandra Thurman, Mimi Kiser, and Laura Beres CHAPTER 6 - DIFFERENCES IN THE PRIVATE COST OF HEALTH CARE BETWEEN PROVIDERS AND SATISFACTION WITH SERVICES: RESULTS FOR SUB-SAHARAN COUNTRIES .……………………………………….…………..…..91 Clarence Tsimpo and Quentin Wodon CHAPTER 7 - MAKING QUALITY CARE AFFORDABLE FOR THE POOR: FAITH- INSPIRED HEALTH FACILITIES IN BURKINA FASO .…………….…………….102 Regina Gemignani, Clarence Tsimpo and Quentin Wodon CHAPTER 8 - BENEFIT INCIDENCE OF PUBLIC HEALTH SPENDING: COMPARING PUBLIC AND FAITH-INSPIRED PROVIDERS .………………..….121 Harold Coulombe and Quentin Wodon iv Acknowledgments This work benefitted from funding from the World Bank as part of a work program on the role of private providers of education and health services. Special thanks are due to Robin Horn, Nicole Klingen, and Rakesh Nangia for supporting this research, as well as to Daniel Cotlear, Frank Dimmock, Mattias Lundberg, Rosalia Rodriguez Garcia, Halsey Rogers, and Karima Saleh for serving as peer reviewers. The editors are grateful to the authors for their contributions towards this collaborative collection, and to Martin Mpungu Lutalo and Erika Yanick for their help in producing this volume. v MAPPING, COST, AND REACH TO THE POOR OF FAITH- INSPIRED HEALTH CARE PROVIDERS IN SUB-SAHARAN AFRICA: A BRIEF OVERVIEW Jill Olivier and Quentin Wodon University of Cape Town and World Bank INTRODUCTION This is the third and final collection in a three-volume series on strengthening the evidence for faith-inspired engagement in health in sub-Saharan Africa. This final volume discusses various ways in which faith-inspired providers can be ‘mapped’, including in terms of whether they are reaching the poor or are located in poor areas. For the last decade one of the main arguments about faith-inspired healthcare and the contribution of religion to health more broadly has been that these providers are not ‘on the map’. This has resulted in many calls at an international level for more ‘mapping’ (Olivier 2010). Karpf (then at the WHO) noted in a meeting on Mapping Standards for Faith-based Providers in Geneva in 2010 that “…the truth is, if you are not on the map, then in the eyes of the donor and Member states, you do not exist� (WHO-CIFA 2010) – and this argument was utilized as a basis for increased mapping efforts on the part of many faith-inspired institutions and their research partners. Progress towards mapping faith-inspired providers has been achieved in recent years, as illustrated by mapping studies illustrated in this volume, such as the study conducted in Zambia and Lesotho in 2005-6 by ARHAP for the WHO. There has been a rise in the use of mapping technologies in this field as in others, including mapping through the use of geographic information systems. At the same time it is important to note that ‘mapping’ is often undertaken and understood in very different ways, with studies ranging from ‘harder’ geographic mapping to ‘softer’ interpretative mapping methodologies. Consider the following example of initiatives broadly related to mapping approaches. In 2010 the World Health Organization adapted the questionnaire of its Service Availability Mapping Survey in order to include faith-inspired categories. This is a form of ‘facilities mapping’ utilizing geographic information systems and health services data – and as a result, faith-inspired providers should slowly become more visible as new surveys are undertaken. Separately, several denominations have recently undertaken large international ‘mapping’ exercises of their health and development assets and programs – as is the case of the Roman Catholic Church and the Anglican Health Desk among others. Some limited service mapping has already been undertaken elsewhere. For example Todd et al (2009) mapped nurses and midwives available to support antenatal care in faith- based health facilities in one region of Tanzania – and provided some training for human- resource-related geographic information systems for the Christian Social Services Commission to utilize this technology. There has also been an increase in emergency (or disaster-relief) mapping, with faith-inspired providers engaged in such collaborative mapping exercises which increasingly use open-source mapping tools. One example is 1 that of the Interagency Haiti Health Facilities Workgroup (coordinated by the US Health and Human Services desk in 2010). There has also been an increase in various types of ‘religious health asset mapping’ (see Olivier et al 2012, Blevins et al 2012), which tend to involve more participatory styles of mapping and focus in part on community interpretations of faith, and faith-inspired influence on health and development. There has also been a rapid increase in literature- based reviews, which are also often labelled as ‘mapping’ - but which are more related to ‘scoping reviews’. Finally, there has been an increase in work aiming to assess the extent to which faith-inspired providers are located in poor areas or serve the poor. This work is typically based on household survey data, as demonstrated by several papers in this collection (see also Wodon, 2013). One example is the application of ‘poverty maps’ techniques to assess the location of health care facilities in order to better understand to what extent faith-inspired facilities are located in poor areas. Another approach relies on household surveys to measure who benefits from services provided by faith-inspired providers. All these various approaches to ‘mapping’ have their own limits and rewards. Clearly these different forms of ‘mapping’ result in different kinds of evidence. In a way, the specific mapping tools are less important than the intention behind the mapping, the arguments being made, the items that are being chosen to be placed on the map, on whose map these items are placed, and the forms of evidence that results. Mapping studies based on ‘hard’ evidence (such as health facilities surveys) are usually valued more highly at a policy level than the ‘softer’ approaches which might seek to understand how individuals and communities perceive or relate to the facilities and assets being mapped. But ‘softer’ approaches tend to result in more nuanced layers of evidence and results, which may also be beneficial. Choosing the right mapping approach depends on the objective sought. OVERVIEW OF VOLUME THREE This collection includes papers from a broad range of researchers and practitioners from the ‘North’ as well as the ‘South’. What is shared is a common interest in uncovering what might be distinctive about faith-inspired health initiatives and institutions, and how they might be placed ‘on the map’. We use the term ‘mapping’ in a broad sense – including research on reach to the poor, and we include studies that move beyond the development of better maps towards considerations of how individuals perceive and engage with the mapped facilities. Therefore, ‘mapping’ means more than the depiction of geographic placement and it deals with ways of recognizing the varied contribution of different kinds of religiously-influenced or faith-affiliated assets, which directly and indirectly have an impact on health. While we focus on Africa, the questions that are raised are likely to be of interest for other regions of the world as well. As in previous volumes, the authors of the various chapters rely on different kinds of research strategies and perspectives. Some papers are full-length analytical articles reporting on new evidence, while others are shorter ‘notes’ summarizing recent work. 2 The collection consists of eight papers. It starts with two introductory chapters, one on the reach to the poor of faith-inspired providers, and the other on mapping frameworks to both document and better understand faith-inspired interventions. Chapter one by Olivier, Tsimpo and Wodon asks whether faith-inspired health care providers in Africa reach the poor more than other providers. Faith-inspired institutions (FIIs) commonly have as their stated mission a desire to provide quality health services to all, and in particular a commitment to serve the poor, for example, by providing services in remote areas where there are none, or by making services more affordable for those in need. Yet it is unclear whether they are able to fulfil this commitment in the current contexts in which they operate – for example by serving the poor proportionately more than other (wealthier) households, or being utilized by the poor more than other providers. Using data from 14 recent nationally representative household surveys in Africa, the paper suggests that when compared with public providers, FIIs on average currently tend to serve the poor slightly less than other population groups. The data also suggest that on average, beyond differences between countries, FIIs do not serve the poor proportionately much more than public providers (the most relevant comparison, given that non-religious for-profit private providers tend to be more oriented towards serving wealthier groups). This does not mean that FIIs do not make special efforts to reach the poor, for example by subsidizing them in order to make services more affordable. However, the data suggest that in current African health contexts, FIIs may no longer be that different from public providers in terms of the clientele they serve. Subsequent chapters in the collection come back to this topic using country case studies, but this chapter sets the overall evidence for the region. The next chapter by Olivier and Wodon is entitled ‘Layers of evidence: Discourses and typologies of faith-inspired community responses to HIV/AIDS in Africa’. The paper has two objectives. The first is to provide a review of the discourses about the religious response to HIV/AIDS in Africa that have emerged from the recent literature, how these discourses has changed over time (from religiophobia to a cautious recognition of the comparative value of faith-inspired interventions), and the conflicting typologies of faith- inspired initiatives that they have inspired. Noting the limits of the existing typologies, the second objective is to suggest conceptually some of the ways in which typologies could be combined in order to be made more useful from an operational point of view. While the analysis is focused on HIV/AIDS, it has broader relevance for faith-inspired health engagement in Africa. The chapter emphasizes the fact that religious institutions and community initiatives are by nature highly diverse and complex – making mapping efforts difficult. The challenge is therefore to work towards multidisciplinary methods of describing and assessing such complexity in a way that is not only conscious of the intricate collaborative environment in which interventions and evidence-gathering take place, but also does not weaken the initiatives we are seeking to support through ill- fitting classification or policy-level paralysis. While improving typologies in this area will remain messy and imperfect work, it appears to be worth the continued effort – and although discussions of evidence, typologies and classifications might seem technical or even abstract, at their heart is a basic understanding that there is potential and opportunity here for positive change. 3 After these two broad introductory chapters, most of the other chapters are devoted to case studies related either to mapping faith-inspired interventions or reach to the poor. Chapter three by Olivier, Cochrane and de Gruchy summarizes the key results of a 2006 mapping study that took place in Zambia and Lesotho. The study aimed to more systematically understand and describe what role religion had in the health-seeking strategies of local communities, and to map both the formal and informal HIV and AIDS initiatives in the process. That the World Health Organization (WHO) would commission such research was unusual given long-standing and deep political sensitivities about religion. Also unusual was the deliberate extension of the study beyond standard quantitative geographic information system (GIS) mapping, towards a participatory community engagement model that qualitatively sought out local perspectives on the value and function of the mapped entities for the communities themselves. In the chapter, the authors address key lessons learned about participatory mapping of religious community responses to HIV and AIDS, focusing on the challenges and strategies of mapping ‘informal’ community-level responses and initiatives. Next, Coulombe and Wodon use data from Ghana to assess the extent to which faith- inspired facilities in that country are located in poor areas. The paper focuses on the facilities associated with the Christian Health Association of Ghana, the largest federation of faith-inspired provider of health services in the country. With 168 hospitals and clinics at the time the paper was written, 70 percent of which are affiliated with the Catholic Church, the federation accounts for more than a fourth of all hospital beds in the country according to administrative data from the Ministry of Health. Using poverty mapping techniques and simple regression analysis, the authors answer the following question: Are CHAG facilities located primarily today in areas that have a high proportion of Catholics or Christians, or in areas that have a high proportion of the population living in poverty? It appears that the location of CHAG facilities is correlated today more with the share of Catholics living in specific districts than with the level of poverty in those districts. In chapter five, Blevins, Thurman, Kiser, and Beres look at the use of mixed mapping methods to uncover community health assets in Nairobi. Their paper summarizes some of the results of a new approach to health asset mapping in Mukuru, an informal settlement in Kenya’s capital city. This particular approach combines a model for community based participatory research with a GIS mapping process. The essay provides background on the rationale for the mapping and of the characteristics of the Mukuru community, a summary of findings and insights gained from the mapping, and a description of the possibilities and challenges of a proposed initiative to mobilize the assets identified as part of a structural HIV prevention initiative. In chapter six, Tsimpo and Wodon note that the issue of whether faith-inspired providers are able to reach the poor depends in part on the cost of the health services provided. Their paper relies on recent nationally representative household surveys for sub-Saharan African countries to assess to what extent the cost of healthcare is a reason for not being satisfied with health services and whether concerns with costs differ between types of providers. The paper also provides estimates of the cost of healthcare in six countries, comparing public, private secular, and faith-inspired providers. The results suggest that 4 cost indeed remains a major concern for households. There are also differences in out-of- pocket costs for households between providers: in many cases public providers being cheaper than faith-inspired providers and private secular providers. Yet the differences depend on the country and are not as large as one might have assumed, given that faith- inspired providers often receive only limited support from the state and are therefore more dependent on cost recovery and other income sources. In chapter seven, Gemignani, Tsimpo and Wodon look at whether quality health care can be made affordable to the poor – using the example of faith-inspired health facilities in Burkina Faso. Based on the results of qualitative fieldwork conducted in 2010, they suggest that a key reason for individuals to seek care in faith-inspired health facilities is the fact that the cost of care is lower than in public facilities (the other reason being that faith-inspired facilities are perceived as providing a higher quality of care). Yet faith- inspired facilities receive only limited support from the state to provide their services. The ability of the facilities to make quality care affordable for the poor is maintained thanks to support in-kind and in cash from religious groups and other donors. This model contributes to the availability of affordable quality care in the communities where the facilities are located, but higher state support for the facilities would help for expansion. Finally, in chapter eight, Coulombe and Wodon rely on the fifth round of the Ghana Living Standards Survey collected in 2005-2006 to conduct a benefit incidence analysis of public spending for health. District-level financial data on public transfers are combined with household survey data on the use of various types of facilities by the population to assess whether public health spending reaches equally various segments of the population. The estimates of benefit incidence are presented separately for public and faith-inspired facilities, given that the latter also benefit from public funding. The analysis suggests substantial variation in transfers and unit costs by districts, with higher costs in areas with the lowest and highest poverty measures, and lower costs in-between. Public health funding is also found to be regressive, in large part because hospitals and clinics still benefited more the better off than the poor, whether they are operated by the government or faith-inspired providers. REFERENCES African Religious Health Assets Programme (ARHAP). 2006. “Appreciating Assets: The Contribution of Religion to Universal Access in Africa.� Cape Town: African Religious Health Assets Programme. Olivier, J. 2010. In Search of Common Ground for Interdisciplinary Collaboration and Communication: Mapping the Cultural Politics of Religion and HIV/AIDS in sub- Saharan Africa. PhD Dissertation, South Africa: University of Cape Town. Todd, S., G. Brubaker, S. Chand, C. Franzén, C. Hafner, A. Kimambo, and P. Pamba. 2009. Human Resources-Geographical Information Systems Data Development and 5 Systems Implementation for the Christian Social Services Commission of Tanzania: Final Report. USAID and The Capacity Project. WHO-CIFA. 2010. NGO Mapping Standards Describing Religious Health Assets. Geneva, Switzerland: Conference supported by the World Health Organization and the Center for Interfaith Action on Global Poverty. Wodon, Q. 2013. Faith and Human Development: Education and Health Services in Africa, New York: Palgrave MacMillan. 6 CHAPTER 1 DO FAITH-INSPIRED HEALTH CARE PROVIDERS IN AFRICA REACH THE POOR MORE THAN OTHER PROVIDERS? Jill Olivier, Clarence Tsimpo and Quentin Wodon University of Cape Town and the World Bank Faith-inspired institutions (FIIs) commonly have as their stated mission a desire to provide quality health services to all, and in particular a commitment to serve the poor, for example, by providing services in remote areas where there are none, or by making services more affordable for those in need. Yet it is unclear whether they are able to fulfil this commitment in the current contexts in which they operate – for example by serving the poor proportionately more than other (wealthier) households, or being utilized by the poor more than other providers. Using data from 14 recent nationally representative household surveys in Africa, this paper suggests that when compared with public providers on a broad macro scale, FIIs currently tend to serve the poor slightly less than other population groups. The data also suggest that on average, beyond differences between countries, FIIs do not serve the poor proportionately more than public providers (the most relevant comparison, given that non-religious for-profit private providers tend to be more oriented towards serving wealthier groups). This does not mean that FIIs do not make special efforts to reach the poor, for example by subsidizing them in order to make services more affordable. However, it suggests that in current African health contexts, FIIs may no longer be that different from public providers in the clientele they serve. INTRODUCTION It is now common to state that faith-inspired institutions (FIIs) contribute a large share of health care services in Africa. Alongside such market share estimates, a strong argument is also made that FIIs have characteristic comparative values, including a preferential option for the poor - in particular, providing services to the ‘rural poor’ where there might otherwise be none. For example, in a World Bank working paper, De Jong (1991) describes health-engaged non-governmental organizations (NGOs) working in Africa. She notes, “…NGOs including missions, involved in health-related activities tend to be particularly represented in poorer, more remote areas, either out of commitment to serve the under privileged (e.g., religious missions often state this explicitly) or because they can fill a gap in such areas not already met by government services…NGO activities may also be concentrated in areas that the government is not serving for political reasons…There is debate, however, on the extent to which NGOs really do reach the poor or the underserved…� More recent examples can be seen in the current documentation of the various Christian Health Associations (CHAs) in Africa. For example, in the 2006 Memorandum of Understanding (MOU) between the Government of Ghana’s Ministry of Health and the Christian Health Association of Ghana (CHAG), a 7 shared purpose is expressed to improve health services in Ghana, “…especially at the rural and deprived communities where CHAG facilities are situated by choice, and experienced in serving such communities, in line with their Christian mission of service to the poor, marginalized and disadvantaged.� Furthermore, the MOU is based on the common principles that, while the Government of Ghana is generally responsible for the provision of health needs of the population as a whole, “CHAG institutions, in line with their Christian teachings, shall target service provision to the poor and the marginalized in the society� (Ghana-MOH and CHAG 2006, more examples below). Unfortunately, while the sentiment that FIIs target the poor in their health and education service provision is shared across Africa and internationally, the evidence of what is happening in practice is rather thin, mostly anecdotal, and at times outdated. This is especially the case for claims that FIIs target the poor in a preferential way - which are made strongly in a wide variety of literature, but for which there appears to be little robust comparative data. The question of whether FIIs in Africa reach the poor preferentially begs a clarification – as there are a number of different questions intertwined within these common perceptions about faith-inspired health services. One question is to what extent FIIs serve the poor ‘proportionally more’ than other population groups internally, that is among their clientele. Another question is whether FIIs serve the poor proportionally in comparison to other providers such as other private or public facilities. It also may be useful to consider whether FIIs employ special strategies to serve the poor (for example, by cross- subsidizing their services – either within a particular facility, or across a system of facilities). It has also been broadly suggested that FIIs serve the poor primarily as a result of their choice of location in remote rural areas where there are no other health providers – begging the question of how their comparative service to the poor is related to their physical location. These are all different and important questions for policies aiming to make health care more accessible and affordable to vulnerable groups, but again the literature to-date has provided only very limited information on each of them. The objective of this paper is to begin to engage with this complex collection of issues by addressing only the first two questions mentioned above. First, is the share of the poor in the clientele of FIIs higher than the share of other household groups? Second, are the services provided by FIIs used by the poor proportionately more than the services provided by other health care providers? To answer these two questions, we rely on data from 14 recent nationally representative household surveys in Africa, with health questionnaire modules sufficiently detailed to permit the identification of faith-inspired providers among the various types of health care providers that households rely upon for their care (see also Wodon 2013). The paper is structured as follows. Section two provides a brief review of the literature, both on the role of FIIs in health care provision in Africa generally, and specifically addressing the understanding that FIIs serve the poor in a preferential way. Section three presents new evidence from the household surveys to look at these questions. 8 REVIEW OF THE LITERATURE Health services and care provided by faith-inspired institutions and communities are ‘significant’ in many countries in Africa. In the colonial period, mission-based hospitals and primary care were a dominant source of healthcare provision in many areas – with colonial governments implementing differently organized health service strategies, some focused on providing services to European employees of the colonial state (primarily in urban areas), and others through Christian missions providing health and educational services to indigenous populations (see Robinson and White 1997, Schmid et al 2008). Different national strategies were adopted after independence as countries dealt with the legacies of their respective colonial administrations. “This included centrally funded health systems, operating out of new national capitals with little local accountability. In time, funding from colonial powers in Europe was replaced by United Nations (UN) agencies and more recently, by other donors� (Schmid et al 2008). In some countries faith-inspired health facilities built in colonial times remained dominant despite attempts at imposed controls of the voluntary sector (such as in Tanzania, Zambia, Kenya and Malawi). Other countries integrated these facilities more fully into public systems, as was the case in South Africa. As to the classic provision of healthcare through ‘church health facilities’, it was bolstered in many countries by a rapid growth of civil society organizations (often supported by international funders) that become engaged in health and development work, creating a complex landscape of non-state, private, or non-profit health provision. Furthermore, in sub-Saharan Africa (SSA), the HIV/AIDS response has had a significant impact on the role of FIIs, with a rapid growth of actors (of the NGO variety) being formed around the year 2000 (see Haddad et al 2008). This particular historical trajectory related in part to the creation of new funding sources contributed to make the involvement of faith-inspired civic institutions in health and health-related service provision especially significant in SSA, and perhaps more so than in any other region in the world. While it may be safe to describe the ‘significance’ of faith-inspired health care, putting a number or estimate to that significance is much more challenging. One challenge is that faith-inspired health services are rarely properly aligned with national health systems, with faith-inspired, private and public health sectors developing in parallel to each other in many countries. The idea that FIIs own a large portion of the health infrastructure has resulted in talk of there being a ‘faith sector’ which is a ‘hidden giant’ in Africa – but this is most often stated with the caveat that integration of FIIs into national health systems and large development programmes is still lacking (see Asante 1998, Schmid et al 2008). In the last two decades there has been a resurgence of interest in non-state and faith- inspired health service provision. This came as a result of several factors, such as growing recognition of the lingering importance of religion to African communities, an increased focus on community-oriented development and health engagement, reforms of national health sectors, an increased focus on the ‘public-private mix’ of health services, 9 and health sectors looking to better harmonize activities and strengthen increasingly fragile and resource-constrained health systems (see Hanson and Berman 1994, Schmid et al 2008). There has also been a burst of interest from governments and international agencies, which have begun to ask if the presence of FIIs can be more clearly demonstrated and mapped, so that they might become stronger allies in the delivery of health services and the accomplishment of global targets – which has generally resulted in increased collaboration (see Marshall and Keough 2005, Olivier and Paterson 2011). However, the evidence as to the role of FIIs in healthcare in Africa remains weak. As Hanson and Berman (1994) put it, “much of what we know is based on anecdotal evidence and ad hoc data collection…There is considerable diversity within the continent, both with respect to levels of expenditure and orientation of health system.� This may be in part as a result of a complex history of secularization and modernization theories and academic tendencies, the ‘faith sector’ became largely invisible, so that while the work of faith-inspired institutions continued, these activities were either hidden from view, or in a few cases, subsumed as part of private, non-state or civil society sectors. As a result, good data are frequently absent, incomplete or stored in protected nodes – for example, with different institutions conducting separate inventories that are often not made publicly available, or utilizing different measures that prevent this information from being comparable and integrated. The evidential landscape is also slanted heavily towards large and organized FIIs with an international or national footprint; towards particular mainstream denominations (with less coordinated faith- inspired groups such as the Pentecostals or ‘traditional’ groupings getting significantly less attention); and towards Anglophone countries which tend to have a larger footprint for FIIs as well as a more developed descriptive literature, written in English (see Schmid et al 2008). The huge diversity of the so-called ‘faith-sector’ also presents many challenges for evidence gathering – with diversity of religious profile across regions, of types of FIIs and services differently engaged in health and development work. As many have noted, this diversity makes any broad regional generalisations dangerous. There is no international inventory or map of faith-inspired health services or facilities, and national and international mapping projects often do not include FIIs who have flown ‘under the radar’ for decades, and remain invisible to national and international views as well as unaligned with national health systems (see ARHAP 2006, Marshall and Van Saanen 2007, Schmid et al 2008, WHO-CIFA 2009). As World Bank President James Wolfensohn (now famously) said in 2002, "half the work in education and health in sub- Saharan Africa is done by the church…but they don't talk to each other, and they don't talk to us" (in Kitchen 2002). In the context of this poor evidence-based, it is still frequently stated that a key ‘comparative advantage’ of FIIs is that they manage to reach the poor - whether this is done by providing services at lower cost for patients, or being located in poor, often rural areas. Examples of statements that combine ‘market share’ estimates with the notion that FIIs primarily serve the poor are given in Table 1 for a subset of countries and for sub- Saharan Africa as a whole. 10 Table 1: Statements on market share and reach to the (rural) poor of FIIs in Africa Country Statement Sources Burundi In Burundi more than one third of health services in rural areas are Hanson and Berman provided by mission clinics 1994, World Bank 1983 Ghana In Ghana while missions provide 25% of total hospital beds in the De Jong 1991, CHAG country, they provide about 46% of beds in the six under- 2006 privileged northern regions; CHAG members cater for an estimated 35-40% of the national population, mainly in the hard to reach rural parts of Ghana Kenya The majority of Christian Health Association of Kenya (CHAK) Muriithi et al 2007 member health facilities are located in rural and remote marginalized areas of the country Malawi The Christian Health Association of Malawi (CHAM) manages Ward et al 2010, 171 health facilities in mainly remote rural areas across the CHAM 2008 country. This makes up 37–40% of all health facilities in Malawi and particularly responds to the need for health facilities in some remote areas with little government coverage Nigeria The Christian Health Association of Nigeria strives to deliver CHAN 2006 healthcare to the furthest and most remote parts of Nigeria (where most member facilities are positioned), reaching out to those who would otherwise not benefit from health care…providing 40% of health services with a special emphasis on the needs of the rural poor Senegal In Senegal, most private sector facilities are located in and around Knowles 1994 Dakar as well as a few large towns … the major exception is the health posts operated by the Catholic Church, most of which are in rural areas Tanzania The Christian Social Services Commission (CSSC) estimates that CSSC 2007, Todd et al FBOs in Tanzania manage 40% of hospitals, 26% of all health 2009 facilities and provide 50% of health services in rural areas Zambia The Christian Health Association of Zambia (CHAZ) accounts for Nussbaum 2005, nearly 30% of Zambia’s total health care provision in general and Mogedal and Steen 50% of rural health care provision 1995, Robinson and White 1998 Zimbabwe Zimbabwe church missions provide 68% of all beds in rural areas; Green and Matthias In Zimbabwe 80%, and in Tanzania 90%, of church hospitals are 2005, Robinson and in rural areas initially less favored by other health care providers White 1998, Gilson et al 1994 Africa (In SSA) health services are often concentrated in urban areas Parry 2003 while rural areas, where most of the population lives, are underserved. Mission hospitals and health care centers are frequently the only such services to be found in these areas Source: Compiled by the authors. All of these examples suggest that FIIs are located in poor and marginalized areas and serve primarily the poor. Many experts (from faith-based and secular positions) are ready to defend such a perspective of FIIs based on their experience in the field. If correct, such assertions should have serious implications for policy, since stakeholders should be more inclined to support FIIs, especially for provision of health care to those with limited access. However, it is difficult to unpack the data on which these statements are based – especially considering the acknowledged dearth of national-scale inventories or maps inclusive of faith-inspired health facilities. There are unfortunately few systematic 11 surveys that can verify such statements, certainly not in a way that could be used to engage with specific policies and strategies that address particular characteristics of faith- inspired service provision. This demonstrates the importance of moving beyond broad statements about the ‘faith sector’ provision in Africa, to evidence-based information that can have a more profound impact at a policy level (see Olivier and Wodon 2012). Said differently, market share estimates that are inclusive of specific indicators which can demonstrate access for the poor and contributions to higher equity in health care seem much more likely to have an immediate impact at a policy level. If these FIIs are indeed the only available facility in specific area, or are targeting a specific poor population, then a different strategy is required for their support. General statements that FIIs in Africa are targeting the ‘rural poor’ are not likely to have much effect without more robust evidence – especially in the complex health service contexts as in Africa. Mission facilities were historically located in both isolated areas, as well as urban centers (see Gilson et al 1994, Schmid et al 2008). Contexts have changed, with some facilities falling into disrepair, and in other locations FIIs have taken up abandoned government facilities in remote areas (Banda et al 2006). McGilvry noted in 1981 that the location of FIIs was often “determined more by ecclesiastical considerations and historical circumstances than by an analysis of health needs. As a result there was frequent overlapping and duplication� – however it is difficult to know whether such ‘ecclesiastical’ considerations still hold sway, or quite what the criteria for the location of FII services currently are. Most African countries now have diverse and migratory populations, where the ‘rural poor’ is just as likely to be located in urban centers. FIIs have also moved and adapted, with many more NGOs (faith-inspired or not) working in areas not served by the traditional mission facilities. It is also suggested that FIIs are more frequently located in conflict areas than other providers, with faith-inspired staff motivated to remain in such contexts as a result of their faith (see Lusey-Gekawaku 2003). Another concern with any broad statement about the preferential option for the poor by FIIs is the always problematic broadness of the category of ‘FIIs’ (or FBOs) – especially given the historic diversity of faith-inspired health providers. For example, a common over-simplification is that all (or even most) mission-based health facilities were historically located in remote rural areas serving only the poor. Of course, many were, but there were also large faith-inspired facilities established in ‘urban’ centers, providing high quality services to all, including the ‘non-poor’. This demonstrates the difficulties in making sweeping statements about all FII’s preferential provision to the poor – especially when such statements are substantiated mainly by pointing to the geographic location of facilities (which is varied), without a better understanding of the nature of clientele served, or the mechanisms employed to subsidize service to the poor. There has also been a rapid expansion over the last twenty years in the number of public facilities operating in poor areas, and to a smaller extent the same trend is also likely to have taken place in at least some countries for non-profit health facilities operated by non-religious NGOs. This is because many countries have significantly expanded the provision of services through health posts or clinics in rural areas, while at the same time 12 the network of facilities operated by FIIs such as the Christian Health Associations have often continued to be oriented towards service provision through larger facilities such as hospitals (which do not reach all poor areas of a country, and are by nature rarely ‘pro- poor’). Thus, the more rapid expansion of non-religious facilities in poor areas may have eroded some of the significance of the location-based edge that FIIs may have had in serving the poor (being reminded that we are considering here the comparative presence of FIIs at a macro-level as considered by policy-makers, rather than the acknowledged value of individual FIIs providing health services in remote and hardship areas). Furthermore, even faith-inspired facilities which were historically established in remote or rural areas, may no longer find themselves serving their originally intended population. The demographic landscape of many countries has changed so much in terms of urbanization and (at least in some countries) poverty reduction in the last 20-30 years that it is no longer clear that the original facilities still operated by FIIs are now mostly located in poor rural areas in many African countries. For example, in a number of countries, FIIs used to be located in coastal areas, but these are the areas that have developed the most in most African countries over the last few decades, and these are also the areas where poverty has been reduced the most. These evolving poverty geographies mean that many FIIs must now reconsider what it means to ‘serve the poor’. Certainly, the broader discourse about FIIs in Africa seems to be stuck in a pattern describing the ‘faith sector’ focus on the ‘rural poor’ – when, in fact, there is a broad re- articulation of this core ethos being enacted. For example, in their annual report of 2006, CHAG demonstrates some of these shifts necessitated by changed circumstances, saying: “CHAG member institutions are located predominantly in the rural areas and are aimed at reaching the marginalized and poorest of the poor. A few are in big towns now but were built there when the towns were small and rural. A few can now also be seen in the slumps of some of the cities. These are targeted at serving the health needs of the poor and vulnerable populations that have been created by urbanization� (CHAG 2006). All such considerations need to be clarified – and the possibility of preferential service to vulnerable groups seems an important place to begin. More provision should also be made for consideration of location, population served, access or availability vis-à-vis other providers. It is not clear whether the ‘significance’ of faith-inspired healthcare is underestimated or overestimated – especially if such issues are not pulled into prominence. Some of these questions may be answered by improved service availability mapping being undertaken by the World Health Organization (WHO) over the next decade. 1 However, at this time, despite a strong stated perception by a wide variety of actors, there is currently little ‘hard evidence’ that overall, FIIs in Africa today reach the poor more than other providers in relation to their geographic location in poor areas. The fact that the provision of healthcare to the rural poor is a core ethos of most FIIs is not what is called into question here. For example, in Uganda it has been noted that: “The faith-base of the (private-not-for-profits) was regarded as a distinguishing 1 The WHO SAMS questionnaire has recently been updated so that it contains a more precise count for faith-inspired facilities and services. The results of these updated surveys are likely to become available over the next 5 to 10 years. 13 characteristic…These values translated…into explicit commitments to serve low-income communities. Thus, some of the agencies assessed by the study included special mention of meeting the needs of the most vulnerable in their mandates� (Schmid et al 2008). Reinikka and Svensson (2010) also argued on the basis of data from a quasi-experiment in the provision of untied block grants to health centers in Uganda, that faith-inspired providers appear not to be motivated by profit or perks maximization, but rather by a desire to make more of their services available and affordable to the poor - that is, they seem to be ‘working for God.’ Uganda is only one example of where there are clear demonstrations of the desire of FIIs to serve the poor. What is still not clear is whether FIIs are able to fulfill this desire to provide preferential option for the poor – in their current contexts which are often influenced by the demands of alignment with national health systems, competition with private for-profit providers, and the concerns around financial sustainability in the resource constrained environments in which they all operate. Indeed, the commitment to serve the most vulnerable has created a core dilemma for FIIs operating in modern health sectors: when their ‘mission’ to serve the poor and marginalized seems to sometimes come into direct conflict with the financial survival of the organization (see Olivier and Wodon 2012). For countries with a very low per-capita income it is extremely difficult to maintain high quality health services that are accessible and affordable for the poor. This faith-motivated ethos is not just a value or organizational characteristic, but impacts directly on financial decisions. For example, many FIIs have developed complex user fee systems based on an ‘ability to pay’ principle - where some patients are charged at a full cost in order to subsidize the healthcare of poorer patients (see Banda et al 2006, Gilson et al 1994). This is a critical time in the history of health-providing FIIs, as they are being pushed to weigh their organizational culture and reason for being, against the realities of financial support and survival, “it is a challenge that cuts to the heart of the religious-health landscape, arguing that if FBOs do have the unique strengths listed above, a ‘value added’, then now is the time to consider just what that value added is ‘worth’ and therefore, in what ways it is to be supported� (Schmid et al 2008) – before it is lost. EVIDENCE ON BENEFIT INCIDENCE At this time, nationally representative household survey data in which households are asked about the type of health care facility they use when seeking care have not been extensively drawn into these discussions about the extent to which FIIs reach the poor. As noted by Olivier and Wodon (2012) in discussing the market share of FIIs, this may be in part because the surveys most frequently used for work on health and development, the Demographic and Health Surveys (DHS) implemented in similar ways in most African countries at regular intervals, do not distinguish between faith-inspired and non- religious providers of care; they only distinguish between public and private providers, often suggesting that private providers provide a large share of all care in Africa, but this does help answer the questions raised above. 14 The fact that household surveys have not been used more extensively for such work is surprising given that this method may gather more robust and nationally representative socio-economic profiles of health service users (these are, for example, the surveys that are used for poverty measurement). For this paper, instead of using DHS data, we assessed the availability of data identifying faith-inspired health providers in the main multi-purpose surveys implemented in approximately 30 African countries. In about half of the surveys that we looked at, there was enough information on the type of provider consulted by households to identify separately public, private non-religious, and private faith-inspired providers. The list of those surveys and countries is provided in table 2. Note that in Ghana, we have two different and independent household surveys at our disposal – we will thus report for that country the average values obtained with the two surveys taken together. Table 2: Selected countries with household surveys identifying FIIs Year of Year of Country (survey name) implementation Country (survey name) implementation Burundi (QUIBB) 2006 Niger (ENBC) 2007 Cameroon (ECAM) 2007 Nigeria (LMS) 2003/04 Chad (ECOSIT2) 2003/04 ROC (QUIBB) 2005 Ghana (CWIQ and GLSS5) 2003 and 2005/06 Senegal (ESPS) 2005 Kenya (KIHBS) 2005 Sierra Leone (SLIHS) 2003 Malawi (HIS-2) 2004 Swaziland (SHIES) 2009 Mali (QUIBB) 2006 Zambia (LCMS IV) 2004 Source: Compiled by the authors. In a separate paper, we used the same data to look at the market share of FIIs – a method which results in significantly lower estimates than is commonly obtained from the comparison of facilities data (Olivier and Wodon 2012). However, in that discussion we also note a possible underestimation of the market share of FIIs in the population as a whole when it is based on household survey data in this way. However, this does not affect the analysis of the extent to which FIIs reach the poor in the same way. Consider the case where faith-inspired providers serve the poor more than other providers in the specific sense that the share of their services obtained by the poor is higher than is the case for other providers. If some faith-inspired facilities are misclassified by households as public facilities in a quasi-random way (the probability of misclassification is similar for all faith-inspired facilities), then the share of the beneficiaries that are poor in faith- inspired facilities would not be affected. As for public providers, the erroneous inclusion of some faith-inspired providers in their pool would lead to a higher share of beneficiaries of public facilities identified as poor than warranted, but the bias should be small because the number of faith-inspired facilities misclassified as public facilities would be small as a proportion of the total number of public facilities. This is because the market share of faith-inspired facilities is still significantly smaller than that of public facilities, especially when one considers non-hospital care, and because only a subset of faith-inspired facilities would be misclassified. In addition, if it turns out that the profile of beneficiaries according to level of well-being is similar between faith-inspired and public facilities, the bias would be even smaller. 15 In addition, as mentioned in the introduction, the question of whether FIIs in Africa reach the poor proportionately more than other providers begs some clarification: proportionally more than other population groups (internally, among their clientele) or proportionally more pro-poor in comparison to other providers such as other private or public facilities? These are really two different questions. In what follows, we rely on the systematic use of the fourteen household surveys mentioned in table 1 to answer the two questions, but we will also make additional reference to other work, including that which is qualitative in nature. Consider first the question of whether internally, among their own clientele, FIIs serve more socio-economically disadvantaged patients. This can be investigated in a partial way in terms of the location of facilities, given that the extent to which FIIs reach the poor is likely to partially depend on this. While we do not do this here, it is worth noting that in the case of Ghana for example, the data suggests that FIIs are not significantly more present in poorer areas than in better-off areas (Coulombe and Wodon 2011). Yet it could be that when relying on the location of facilities, we are not clearly seeing how FIIs might be providing preferential option for the poor, even located in less-poor areas, because of data gaps on who is served by whom within districts. This is where the household survey data becomes useful. In table 3, data are provided as to the share of the users of services provided by FIIs belonging to various quintiles of well-being, with each quintile accounting for twenty percent of the population, from the poorest to the richest. 2 Although poverty estimates vary between countries, in most countries the bottom two or three quintiles can be considered as representing the poor. The evidence from the fourteen countries suggests that for the most part, FIIs do not serve the poor proportionately more than wealthier groups. The data actually suggest that on average the clientele of FIIs tends to be less poor than the population as a whole, probably in large part because many FIIs operate hospitals and clinics, and care provided in these types of facilities tends to be more expensive and less accessible to the poor than care provided in smaller health posts (also, the larger the facility, the more likely it is that it will be located in an urbanized area with lower levels of poverty). In addition, many among the poor simply do not seek care when ill or sick because they cannot afford the cost of care. This means that the share of households who seek care in higher quintiles of well-being is often larger than in lower quintiles. Of course, this does not mean that FIIs do not make special efforts to reach the poor or to make care more affordable for them. But due to the nature of the service that they provide and the broader socio-economic constraints faced by the population when seeking care, even if such efforts may indeed succeed in reaching some among the poor, overall the services provided by the facilities are not typically ‘pro-poor’. There are, however, some differences between countries. For example, Cameroon and Swaziland appear to be countries where the use of facilities operated by FIIs in the top 2 The quintiles are based on measures of consumption per capita or per equivalent adult normalized by poverty lines accounting for differences in cost of living between areas within a country, in order to ensure consistency with poverty measurement techniques. 16 quintile is twice that in the bottom quintile, and in Nigeria, the differences are even larger (although we trust this data for that specific country less due to the low market share observed for FIIs in the survey, which seems to be more at odds with the facilities-based estimates of market share than is the case in the other countries). By contrast, in the Republic of Congo, FIIs seem to serve the poor proportionately more than wealthier groups in the population. Note that in the case of Mali, the pattern is a bit erratic, essentially because the sample size for estimating the benefit incidence of the services provided by FIIs is rather small. If one were to exclude the outlier value of 2.8 percent benefit incidence in the top quintile for that country, the average share of services obtained by the wealthiest quintile for the remaining thirteen countries would be 24.2 percent, similar to what is observed for the fourth quintile. Table 3: Benefit incidence by quintile of well-being of services provided by FIIs (%) Q1 Q5 Q2 Q3 Q4 All (Poorest) (Richest) Burundi 17.6 20.9 18.6 22.5 20.4 100.0 Cameroon 13.9 16.0 19.9 24.0 26.2 100.0 Chad 23.0 22.8 19.5 17.0 17.7 100.0 Ghana 21.5 18.3 19.9 20.0 20.5 100.0 Kenya 12.8 21.9 12.3 28.5 24.5 100.0 Malawi 10.5 15.0 19.9 25.7 28.9 100.0 Mali 27.1 13.1 16.7 40.3 2.8 100.0 Niger 8.9 13.2 49.8 13.5 14.7 100.0 Nigeria 8.0 9.8 11.3 17.7 53.3 100.0 Republic of Congo 31.8 19.5 7.0 22.7 19.1 100.0 Senegal 19.0 11.5 18.1 37.1 14.3 100.0 Sierra Leone 13.4 20.9 13.4 26.2 26.2 100.0 Swaziland 13.2 17.2 15.9 27.6 26.0 100.0 Zambia 21.0 18.5 15.3 22.1 23.1 100.0 Average 17.3 17.0 18.4 24.6 22.7 100.0 Source: Authors’ estimations using household surveys. However, are FIIs comparatively more pro-poor than public or other private facilities? A simple way to answer that question is to compare the market share of FIIs in the various quintiles of well-being as well as in the population as a whole, as done in table 4. Note that in table 4, non-religious private providers are a highly heterogeneous group that includes fairly different types of providers, from for-profit hospitals and clinics which tend to serve the wealthier segments of the population to chemical stores and pharmacists, as well as informal health practitioners such as traditional healers who serve the poor more (among traditional healers, some could be considered as faith-inspired, but that is not an obvious call, and for the specific purpose of this paper which focuses more on assessing facilities-based services provided by FIIs, it seems more appropriate to exclude traditional healers from FIIs). On average, and when looking at the average market share across the fourteen countries provided in the last row of the table (without using population weights to reflect the size of each country and the market share of FIIs in each country), the profile of the market share by quintile of faith-inspired facilities is similar to that of public facilities. There are relatively few differences in market shares between the first four quintiles, but the market share of both faith-inspired and public facilities drops a bit in the top quintile, essentially 17 reflecting the higher use of modern non-religious private health facilities in that group. The market share of FIIs in the poorest quintile is however proportionately slightly higher than is the case in the public sector, but besides that, differences are small. The fact that there are few systematic differences in the market shares of FIIs between quintiles suggests that in comparison to public providers, the reach to the poor of FIIs is somewhat similar. Thus, FIIs serve the poor slightly less than other groups, as was shown in table 3, but this is also the case for public providers on average. When one considers the market share of FIIs in the various quintiles, it does not change much, showing that FIIs do neither much better, nor worse than public providers, but they do a slightly higher market share among the very poor. As before, there are of course differences between countries. In Chad for example, the market share of FIIs among households from the poorest quintile, at 14.8 percent, is twice as large as that among the richest quintile, at 7.0 percent, and there is clear indication that the market share declines once one considers richer segments of the population. Thus, in Chad, FIIs serve the poor proportionately more than other providers. In Malawi, at least according to the survey estimates, the reverse is observed: the market share of FIIs is higher among wealthier household than it is among the poor. In table 4, there are three countries where FIIs tend to serve the poor slightly more than all other providers on average, three countries where the reverse is observed, and in the other countries, the conclusion is typically not clear-cut. The conclusion that emerges from the data in table 4 is that in terms of market share, FIIs do not seem to serving the poor proportionately more than the public sector, except perhaps in the bottom quintile of well-being, but even there the evidence is limited. 18 Table 4: Market share of FIIs among households groups by quintile of well-being (%) Public providers Faith-inspired providers Private non-religious providers Q1 Q2 Q3 Q4 Q5 All Q1 Q2 Q3 Q4 Q5 All Q1 Q2 Q3 Q4 Q5 All Burundi 69.3 69.7 68.0 69.3 70.1 69.3 11.8 14.1 10.7 11.2 10.2 11.5 18.9 16.2 21.4 19.6 19.7 19.2 Cameroon 42.2 42.9 44.2 46.2 47.3 44.9 14.6 13.3 15.1 16.9 15.3 15.1 43.3 43.9 40.7 36.9 37.5 40.0 Chad 45.9 50.6 62.5 58.4 58.7 55.8 14.8 14.4 11.5 8.4 7.0 10.7 39.3 35.1 26.0 33.2 34.3 33.5 Ghana 43.9 43.4 43.6 44.2 43.8 43.8 6.1 4.7 5.4 5.2 5.0 5.2 50.1 52.0 51.1 50.7 51.4 51.1 Kenya 53.3 47.8 53.0 50.2 41.7 49.0 3.1 5.1 2.6 5.4 4.7 4.2 43.6 47.1 44.4 44.4 53.7 46.8 Malawi 40.2 40.9 36.0 34.4 34.5 36.9 2.6 3.0 3.7 4.4 5.5 3.9 57.2 56.2 60.3 61.2 60.1 59.1 Mali 56.1 69.4 72.6 77.4 66.4 69.4 2.1 0.8 1.0 1.7 0.1 1.0 41.8 29.8 26.5 20.9 33.5 29.6 Niger 55.3 44.4 44.1 54.9 52.7 50.3 0.9 1.2 3.6 1.0 0.8 1.5 43.9 54.4 52.3 44.2 46.5 48.2 Nigeria 58.1 53.0 56.5 54.1 42.8 50.2 2.3 1.5 1.1 1.4 2.7 1.9 39.7 45.5 42.4 44.5 54.5 47.9 Republic of Congo 40.4 43.9 44.8 44.7 46.1 44.0 7.1 3.0 4.0 2.7 3.5 4.0 52.5 53.1 51.2 52.6 50.4 52.0 Senegal 61.8 65.0 69.2 68.2 62.1 65.0 2.3 2.1 1.9 4.2 1.3 2.3 36.0 32.9 28.9 27.6 36.6 32.7 Sierra Leone 60.9 64.1 60.8 62.1 57.4 60.1 9.3 8.5 6.7 7.5 3.8 6.1 29.9 27.5 32.5 30.4 38.8 33.8 Swaziland 79.0 75.5 72.3 64.3 52.4 66.4 15.0 12.2 10.7 15.8 12.5 13.2 6.0 12.3 17.1 19.9 35.1 20.5 Zambia 58.5 54.4 59.3 54.2 51.4 55.0 8.9 6.8 5.1 5.9 5.2 6.2 32.7 38.8 35.7 39.9 43.3 38.8 Average 54.6 54.6 56.2 55.9 52.0 54.3 7.2 6.5 5.9 6.5 5.5 6.2 38.2 38.9 37.9 37.6 42.5 39.5 Source: Authors’ estimations using household surveys. 19 When combined with the evidence provided on benefit incidence, it therefore does not seem valid to claim that in general, all FIIs serve primarily the poor, or that they do more so than the public sector, even if they do serve the poor more than other non-state (and non-religious) service providers. In fact, because many FIIs are at least in part privately funded (now often partially subsidized by the state), the fact that they serve the poor more than other non-state, non-religious facilities-based providers is a positive and important achievement which certainly deserves more attention – however this a rarely acknowledged nuance in the literature which broadly claims a special reach to the poor for all FIIs. It should be clear that the analysis presented in this paper has limits, and no simple conclusion about FIIs’ reach to the poor in Africa should be removed from this discussion. Benefit incidence analysis or market share comparisons between providers are not the only indicators that can be relied upon to assess whether FIIs serve the poor in a preferential way. Another indicator could be that of the cost of service – even if FIIs do not serve the poor proportionately more than public providers, and even if they tend to have (slightly) more patients from wealthier than poorer backgrounds, it could be that FIIs make special efforts to make their services more affordable to the poor. There is some partial evidence to that effect for several of the countries detailed above. In Ghana for example, detailed analysis of the private cost for households of health care suggests that FIIs may indeed be subsidizing the cost of care for the poor (Coulombe and Wodon 2011). In Burkina Faso, on the basis of in-depth qualitative work in six clinics, Gemignani et al (2011) suggest that a key reason for individuals to seek care in FIIs is the fact that the cost of care is lower than in public facilities (the other reason being that faith-inspired facilities appear to provide better quality of care). The ability of FIIs to make good quality care affordable for the poor is maintained in the clinics through support in kind and in cash from religious groups and donors. Still, considering in a systematic way the ways in which FIIs may manage to make their services more affordable to the poor is beyond the scope of this paper, and before making any general statement on cost and affordability, more research is needed. CONCLUSION There is substantial diversity today in how FIIs provide care in African countries, and whom they serve. FIIs might often aim to serve all – but they also usually have an explicit commitment to providing preferential option for the poor and the vulnerable. The extent to which they are actually able to do so in their current resource-constrained environments remains an open question. What household survey data suggests is that, as is the case with other facilities-based providers, FIIs actually tend to serve the poor less than other population groups, most likely in part because cost (and in some cases distances to facilities) still represents major obstacle to care for the poor. The data also suggest that on average, although there are differences between countries, FIIs do not serve the poor proportionately more than other providers, or at least than public providers (given that, as expected, non-religious for-profit private providers tend to be more oriented towards wealthier groups). This still does not mean that FIIs do not make special efforts to reach the poor, for example by subsidizing them in order to make services more 20 affordable for them. But beyond anecdotal evidence, systematic data to back up that specific claim is still not readily available. These results, which point to a more limited reach to the poor among FIIs than is commonly stated in the broad literature, should not cast doubt on the intention or desire of many FIIs to reach the poor. But FIIs face a number of constraints, in terms of their location, cost recovery mechanisms, and in many cases rules governing national health systems. As FIIs are becoming more integrated in such systems, a difficult choice may have to be confronted. Whether FIIs should consolidate their current services, or whether they should (and can afford to) institute changes to direct their energies to again prioritize the poor is unclear. This type of decision is also linked to other important decisions that have to be made by FIIs, for example, prioritizing primary health care versus facilities- based care. In any case, such reflections challenge us to push beyond broad, advocacy- oriented statements as to whether FIIs in Africa reach the poor proportionately more than others – towards more operational and practical questions, such as what could be done to help FIIs fulfil their desire to serve the poor more. In addition to statistical evidence, more research is needed in order to better document and learn from the multiple initiatives that are being enacted by FIIs to better their reach to the poor. This is especially important if a case is to be made that FIIs require additional or special support from national and international policy-makers. Again, as public health services are (in some places) being expanded into more poor areas and FIIs are often increasingly incorporated in national health systems, these questions and initiatives become critical. What does a core commitment to the poor mean (operationally) in the context of our current health systems? In settings where FIIs play an especially important role, such as in fragile states with very low per-capita income levels, how can they maintain high quality health services that are accessible and affordable for the poor? What data and evidence needs to be gathered (by the FIIs themselves and others) to demonstrate these priorities? Such questions should drive future investigations. REFERENCES African Health Assets Programme (ARHAP). 2006. Appreciating Assets: The Contribution of Religion to Universal Access in Africa. Cape Town: African Religious Health Assets Programme. Asante, R. K. O. 1998. Sustainability of Church Hospitals in Developing Countries: A Search for Criteria for Success. Geneva: (WCC) World Council of Churches, 1998. Banda, M., E. Ombaka, S. Logez, and M. Everard. 2006. Multi-Country Study of Medicine Supply and Distribution Activities of Faith-Based Organizations in sub- Saharan African Countries. Geneva: World Health Organization and Ecumenical Pharmaceutical Network. 21 Christian Health Association of Ghana (CHAG). 2006. “Annual Report: June 2005 - May 2006.� Accra. Christian Health Association of Malawi (CHAM). 2008. “HMIS Status Report.�Lilongwe. . Christian Health Association of Nigeria (CHAN). 2007. Christian Health Association of Nigeria website.. http://channigeria.org/aboutus.html [Accessed 15/08/2011]. Coulombe, H., and Q. Wodon. 2012a. “Mapping Religious Health Assets: Are Faith- Inspired Facilities Located in Poor Areas in Ghana?� In J. Olivier and Q. Wodon, eds, Mapping, Cost, and Reach to the Poor of Faith-inspired Health Care Providers in sub- Saharan Africa. HNP Discussion Papers, World Bank, Washington, DC. Coulombe, H., and Wodon Q. 2012b. “Benefit Incidence of Public Health Spending in Ghana: Comparing Public and Faith-Inspired Providers.� In J. Olivier and Q. Wodon, eds, Mapping, Cost, and Reach to the Poor of Faith-inspired Health Care Providers in sub-Saharan Africa. HNP Discussion Papers, World Bank, Washington, DC. CSSC. 2007. “Christian Health Associations at Crossroad Toward Achieving Health Millenium Development Goals.� Christian Health Association’s Conference, 16-18 January 2007, Dar es Salaam, Tanzania: Christian Social Services Commission. De Jong, J. 1991. “Nongovernmental Organizations and Health Delivery in Sub-Saharan Africa.� World Bank Working Paper, Population and Human Resources Department, World Bank, Washington, DC. Gemignani, R., Q. Wodon, and C. Tsimpo. 2012. “Making Quality Care Affordable for the Poor: Faith-Inspired Health Facilities in Burkina Faso.� In J. Olivier and Q. Wodon, eds, Mapping, Cost, and Reach to the Poor of Faith-Inspired Health Care Providers in sub-Saharan Africa. HNP Discussion Papers, World Bank, Washington, DC. Ghana-MOH and CHAG. 2006. “Memorandum of Understanding and Administrative Instructions.� Accra: Ministry of Health and Christian Health Association of Ghana. Gilson, L, P. D. Sen, S. Mohammed, and P. Mujinja. 1994. “The Potential of Health Sector Nongovernmental Organizations - Policy Options.� Health Policy and Planning 9(1): 14-24. Green, A., and A. Matthias. 2005. “Where do NGOs Fit In? Developing a Policy Framework for the Health Sector.� Development in Practice 5(4): 313-323. Haddad, B., J. Olivier, and S. De Gruchy. 2008. The Potential and Perils of Partnership: Christian Religious Entities and Collaborative Stakeholders Responding to HIV and AIDS in Kenya, Malawi and the DRC. Cape Town: Africa Religious Health Assets Programme. 22 Hanson, K., and P. Berman. 1994. “Non-Government Financing and Provision of Health Services in Africa: A Background Paper.� Washington, DC: United States Agency for International Development. Kitchen, M. 2002. “World Must Coordinate Efforts, end Waste, says Wolfensohn.� UN Wire http://www.unwire.org/unwire/20021025/29832_story.asp. Knowles, J. C., A. S. Yazbeck, and S. Brewster. 1994. The Private Sector Delivery of Healthcare: Senegal. Bethesda, MD: Health Financing and Sustainability (HFS) Project. Lusey-Gekawaku, H. 2003. The Churches Confronted with the Problem of HIV/AIDS: Analysis of the Situation in Five Countries of Central Africa. Geneva: World Council of Churches. Marshall, K. and M. Van Saanen. 2007. Development and Faith: Where Mind Heart and Soul Work Together. World Bank, Washington, DC. Marshall, K, and L. Keough. 2005. Finding Global Balance: Common Ground Between the Worlds of Development and Faith. World Bank, Washington, DC. McGilvray, J. 1981. The Quest for Health and Wholeness. Tübingen: German Institute for Medical Mission. Mogedal, S., and S. H. Steen. 1995. “Health Sector Reform and Organizational Issues at the Local Level: Lessons From Selected African Countries.� Journal of International Development 7 (3):349-67. Muriithi, P., N. Munguti, R. Ayah, and D. Ongore. 2007. “A situational Analysis Study of the Faith-Based Health Services Vis-a-vis the Government Health Services.� Nairobi, Kenya: Ministry of Health Republic of Kenya, Christian Health Association of Kenya and German Technical Cooperation. Nussbaum, S., editor. 2005. The Contribution of Christian Congregations to the Battle with HIV and AIDS at the Community Level. Oxford: Global Mapping International. Olivier, J., and G. Paterson. 2011. “Religion and Medicine in the Context of HIV and AIDS: A Landscaping Review.� In B. Haddad, editor, Religion and HIV/AIDS: Charting the Terrain, Pietermaritzburg: Cluster Publications. Olivier, J., and Q. Wodon. 2012. “Playing Broken Telephone: Assessing Faith-Inspired Health Care Provision in Africa.� Development in Practice, 22(5-6): 819-34. _____. 2012. “Market Share of Faith-Inspired Health Care Providers in Africa: Comparing Facilities and Multi-Purpose Integrated Household Survey Data.� In J. Olivier and Q. Wodon, eds, The Role of Faith-Inspired Health Care Providers in sub-Saharan 23 Africa and Public-Private Partnerships, HNP Discussion Papers, World Bank, Washington DC. Parry, S. 2003. Responses of the Faith-Based Organisations to HIV/AIDS in sub-Saharan Africa. Geneva: World Council of Churches and Ecumenical HIV/AIDS Initiative in Africa. Reinikka, R., and J. Svensson. 2010. “Working for God? Evidence from a Change in Financing of Not-For-Profit Health Care Providers in Uganda.� Journal of European Economic Association 8(6):1159-1178. Todd, S., G. Brubaker, S. Chand, C. Franzén, C. Hafner, A. Kimambo, and P. Pamba. 2009. Human Resources-Geographical Information Systems Data Development and Systems Implementation for the Christian Social Services Commission of Tanzania: Final Report. Washington, DC: USAID and The Capacity Project. Ward, N., J. Kaybryn, and K. Akinola. 2010. Faith in the System: The Impact of Local HIV Responses on Strengthening Health Systems in Malawi and Chad. Teddington, UK: Tearfund. WHO-CIFA. 2009. “Report on WHO-CIFA Consultation: NGO Mapping Standards Describing Religious Health Assets.� 10-12 November 2009, Chateau de Bossey, Bogis- Bossey, Switzerland. World Bank. 1983. “Country Population and Health Sector Reviews (1983–1992).� Unpublished Country Memoranda, Washington, DC. Wodon, Q. 2013. Faith and Human Development: Education and Health Services in Africa, New York: Palgrave MacMillan. 24 CHAPTER 2 LAYERS OF EVIDENCE: DISCOURSES AND TYPOLOGIES OF FAITH-INSPIRED COMMUNITY RESPONSES TO HIV/AIDS IN AFRICA Jill Olivier and Quentin Wodon University of Cape Town and The World Bank This paper has two objectives. The first is to provide a review of the discourses about the religious response to HIV/AIDS in Africa that have emerged from the recent literature, how these discourses has changed over time (from religiophobia to a cautious recognition of the comparative value of faith-inspired interventions), and the conflicting typologies of faith-inspired initiatives that they have inspired. Noting the limits of the existing typologies, the second objective is to suggest conceptually some of the ways in which typologies could be combined in order to be made more useful from an operational point of view. INTRODUCTION Engagement with communities has become a cornerstone of the international response to HIV/AIDS. There has not only been an instrumental recognition of the importance of communities in implementing of specific programs and policies, but also a recognition of the need to understand better the way in which local communities act and engage with their contextualized epidemic through local responses and initiatives. However, our knowledge of what community initiatives responding to HIV/AIDS look like, how they behave or what impact they have is still limited. Our tools and methods for understanding and supporting community initiatives remain inadequate in large part due to a limited evidential basis with which to work. A particular question has arisen as to the nature, role and impact of religious communities and community initiatives. 3 There has been greatly increased advocacy for engagement with religious communities. Dialogue between religious leaders, religious institutions, multi-laterals and others working at a policy level has also increased over the last two decades. As a result, there is now an emerging literature addressing the ‘religious response to HIV/AIDS in Africa’ – some of which deals with community-level responses, albeit through a variety of lenses. Within both this literature, and the broader policy-level advocacy work, much has been said about the ‘comparative value’ of religious communities, initiatives and institutions – with arguments being made that religious communities have a particular or characteristic impact on HIV/AIDS (as well as health and development more broadly) and that the ‘faith sector’ contains untapped assets or potential which have not been properly leveraged in the pandemic response. 3 We use ‘religion’, ‘religious entity’ and ‘initiative’ in this article as broad terms - recognizing the inadequacy and lack of precision of the terminology. 25 However, while the international policy environment may have moved towards a more interested attitude towards religious communities and initiatives (in sharp contrast to earlier periods when religious activities were seen as bringing mainly negative value), this increased attention and literature has not necessarily resulted in any significant policy or strategies for a particular kind of engagement with religious communities and initiatives that takes into account such special characteristics. Said differently, while there are many more policy documents and reports recognizing the importance of religious communities – there are few targeted operational strategies that can be acted upon. This lack of clarity is largely as a result of a failure of evidence, both due to significant knowledge gaps, and also as a result of clashing frames and typologies being applied to the assessment of religious community initiatives responding to HIV/AIDS. In this context, this paper has two main objectives. The first is to provide a broad overview of these main perspectives or ‘discourses’ about the religious response to HIV/AIDS in Africa that have emerged out of the international research and policy environs: how these discourses have changed over time in the last three decades, and the typologies of faith-inspired community initiatives that they have inspired. 4 This provides us with a review of dominant perspectives visible in the literature which have driven policy and research action. Thereafter, noting the limits of the existing typologies used in the literature, we suggest in a conceptual and admittedly rather introductory fashion some of the ways in which these typologies could be combined and made more useful from an operational point of view – and for utilization at a policy level. In so doing, we make a few recommendations for a more properly integrated approach to assessing the religious community response and a return to some of the more basic forms of evidence-gathering that may have greater relevance for policy-level collaborative action and frameworks. The paper is structured as follows. Section two reviews how attitudes towards the religious community response have changed over time. We identify three stages in this evolution, starting with the perception at the early stages of the HIV/AIDS pandemic that religion is at best irrelevant, or may even bring negative value to the HIV/AIDS response. Next comes a turn-around as of the mid to late 1990s in which much more enthusiasm is seen about the potential comparative advantages of religious health assets, but without actual evidence as to their performance. Finally, we review today’s somewhat more cautious mixed bag of opinions and calls for bridging the evidence gap. Next, in section three, we discuss some of the main typologies that have been used to frame the religious community response, in terms of the form of religious entities, their function, or their religiosity. Recognizing the well-acknowledged limits of existing typologies, we then explore in section four ways in which they could be combined to be more useful for policy makers. 4 A broader analysis of this literature can be seen in Olivier (2010). Many examples provided in this article are taken from four religious community mapping studies implemented in eight sub-Saharan African countries by the African Religious Health Assets Programme (see ARHAP 2006, Haddad et al 2008, Schmid et al 2008, Thomas et al 2006). 26 ATTITUDES TOWARDS THE RELIGIOUS COMMUNITY RESPONSE In 1999, Garner wrote an article, asking the question “religion in the AIDS crisis: irrelevance, adversary or ally?� Although more than a decade has passed since he asked this question, it is still difficult to provide a concise answer. There has been a great flurry of debate as to what the comparative value (also described in the literature as ‘comparative advantage’ or ‘value-added’) of religious communities entails, and whether such initiatives and institutions have resulted in improved health and development - but little clarity on what strategic engagement with- or leverage of these might look like. It is helpful to start our discussion of this field by briefly considering some of the key phases the international research and policy environment has gone through – especially with regard to this question of the comparative value of HIV/AIDS-engaged religious communities and community initiatives. We do not systematically detail here the literature on the religious response to HIV/AIDS (of which there are several reviews newly available, see for example Haddad et al 2011), but rather provide a broadly chronological assessment of the dominant themes and attitudes towards religious community engagement as is visible in the HIV/AIDS literature and policy environment (Olivier 2010). The early stage: Religion as irrelevant or of negative value In the first few years of the HIV/AIDS pandemic, international inquiry into the disease- complex was largely blind to religion and religious entities (REs) working in development and health. Under the influence of secularization and modernization, religious involvement in development had become invisible to the majority of scholars and international policy-makers who had become largely ‘religion-blind’. 5 Furthermore, modernization held (and holds) an inherent bias towards religion, where religions were seen as obstacles to progress (Melkote and Steeves 2001). In the early stages of the pandemic, when the biomedical perspective dominated the inquiry (Weeks 1989, Plummer 1988) there was little mention made of religion or of REs. Of course, many REs were engaged in HIV/AIDS work and partnering with international organisations and governments from the very beginning stages. 6 But in terms of categorization, REs were generally assumed to be part of civil society, clustered as NGOs and community-based organizations (CBOs); or even more generally with non- state health service providers - with religion seen simply as one of many culturally relevant variables. Liebowitz (2002) suggests a number of factors which resulted in a belated consideration of the potential comparative advantages of religious engagement in the HIV/AIDS agenda including: the complex nature of the topic, the surrounding controversy, and the relatively modest impact of religious leaders in the developed world from where most of the research agendas are driven. 5 The secularization thesis suggests, among other things, that as societies develop and modernize they will tend to follow the pattern that developed in Europe during the 20th century where religion has been increasingly relegated to the back seat and seemed to lose its influence in public life. Few religious scholars nowadays defend this view. 6 Certainly there are many noted cases of a very early response by established health-engaged REs in Zambia, Lesotho, Kenya and Malawi for example (ARHAP 2006, Haddad et al 2008). 27 The first major appearance of religion in both the international and Africa-focused HIV/AIDS materials comes several years later, during what Weeks (1989) calls the phases of moral panic and crisis management, when HIV/AIDS began to be described as a ‘gay plague’ (in the USA and internationally), and the marginalisation of risk groups developed into a moral panic. Although Weeks is addressing the North American context, it is relevant that this phase of moral panic brought a specific discourse about religion into the international inquiry, and it is clear that this moral discourse was swiftly visible in materials addressing the African HIV/AIDS pandemic. Seidel (1993) describes the emergence of powerful medico-moral discourses, which she sees as being of primarily Christian intervention and frequently judgemental (such as those representing AIDS as God’s punishment). These medico-moral discourses show a complex weaving of religious and biomedical ‘truths’ as people struggled to understand this devastating phenomenon, often to disastrous effects. In reaction against these moral discourses, critical commentary arose that took a strongly adversarial position against religion, describing religion and REs as a barrier to effective HIV/AIDS intervention. Stories of religious leaders shunning PLWHA grew rapidly at this time, and the inquiring gaze towards religion and HIV/AIDS frequently becomes narrow-eyed with suspicion. As Green (2003) described, “During the early years of the HIV/AIDS pandemic, many people who worked in HIV prevention believed religious leaders and organisations were intrinsically antagonistic to what they were trying to accomplish.� UNAIDS took it a step further and named opposition from religious authorities as “perhaps the greatest obstacle to AIDS prevention activities in many countries� (Pisani 1999). Significantly, in the 1990s, the main focus of the international inquiry shifted from a ‘gay disease’ to an ‘African plague’ (Treichler 1992), as HIV/AIDS turned into an administratively chronic disease in most Western countries, but continued to grow exponentially in African states as an epidemic disaster and disease of development (Fox 1992, Rosenberg 1992). Negative attitudes towards religion and REs were woven into such discourses of African AIDS, and narratives of stigmatising male African religious leaders began to flourish, although with little underlying evidence-base to assess the impact or scope of these kinds of behaviours. ‘Religionophobia’ in the form of doubt, suspicion and negative perceptions of religion and religious involvement in development and health continues to linger today (Cochrane 2008, Marshall 2009), and is evident in a number of areas of HIV/AIDS inquiry, tied to the emergence of religion into public life as a liability (particularly discourses about terrorism and religious fundamentalism), and continuing discourses of secularisation and modernisation. Concerns and frustrations also linger, for example, towards the detrimental effect of some religious leaders to HIV prevention strategies, the fear that religious organisations may use public funds for proselytizing or, “concern that ideological considerations are replacing sound empirical evidence of effectiveness in delivering health services� (Breger in Woldehanna et al 2005). All this fed into a dominant perspective of a negative or detrimental religious response to HIV/AIDS in Africa, a response that has been comparatively worse than the ‘secular’ response. 28 The turn-around: Cautious enthusiasm without systematic evidence Things started to turn around late in the twentieth century. While the HIV/AIDS pandemic hit its stride in Africa, contrary to the predictions of modernist and secular theories, religion began to re-emerge into public life on a number of startling fronts (Cochrane 2003, Derrida and Vattimo 1998). 7 Religious movements across the globe have been flourishing, for example, Pentecostal and charismatic varieties of Christianity in sub-Saharan Africa, a highly political Islam in North Africa, and religious revivals in Asia and the United States (Ellis and Ter Haar 2001). Religion has become increasingly important in the political sphere, where the ‘resurgence of religion’ has been treated both with welcome and alarm, context and perspective depending (Asad 2003). Less public, but as influential has been the re-emergence of religion in research and scholarship. The most obvious example of this has been in the public critique (or failure) of the secularisation and modernism theses, which a range of commentators and scholars began noting had critical flaws, in particular the basic idea that development or modernisation equals secularisation. Berger, a key proponent of the secularisation thesis, bravely admits in 1999 that, “…the assumption that we live in a secularised world is false. The world today…is as furiously religious as it ever was, and in some places more so than ever. This means that a whole body of literature by historians and social scientists loosely labelled ‘secularisation theory’ is essentially mistaken. In my early work I contributed to this literature.� This is not to say that religion has re-emerged into academia and scholarship in a blindly positivistic way; to the contrary, it has emerged into a deeply suspicious and antagonistic environment, but what is clear is that it cannot be ignored. Tied to this broader reemergence of religion into public life and scholarship, researchers and policy makers began increasingly to turn their gaze towards the so-called ‘faith sector’ in relation to HIV/AIDS response. In Africa this interest came in a broader context of generally failing health systems and on the heels of what appeared to be unsuccessful attempts at health and development policy reform. HIV/AIDS itself further undermined confidence in existing development and public health strategies (Farmer 1999, Kim et al 2000). In 2000 World Health Organisation (WHO) Director General Brundtland stated that to confront the diseases of poverty, “we must strengthen health systems. We must also go beyond the traditional health sector - working with people in their homes, their work places, their schools, their community halls and their places of worship.� The literature of the late 1990s and beyond shows two main reasons given for the renewed interest in REs: firstly, the recognition of the importance of religion to individual and community behaviour and decision-making, arguing for intervention strategies that are cognisant of that (Benn 2002) and secondly, the possibility of assets 7 As noted by Ellis and Ter Haar (2001), “Labelling it a ‘revival’ or ‘resurgence’ in some ways conveys the misleading impression of a trend that previously existed but that had gone underground, whereas in fact many of today’s most visible religious movements have long been publicly active, but have only recently become subject to academic scrutiny.� 29 held by religious communities that could be leveraged through partnership in a context of limited resources (Haddad et al 2008, Olivier et al 2006). The first reason was powered by the general failure of many HIV/AIDS interventions that were shown to be culturally inappropriate and the belated realisation that religion is important to the lives and decisions of the people most affected by HIV/AIDS. The second reason given for this renewed interest, that religious communities might hold assets for partnership, can be seen to have roots in broader trends in the HIV/AIDS agenda, to now treat HIV/AIDS as a disease of development and society, rather than a medical problem, and a pandemic primarily located in resource-poor settings characterised however by booming civil society sectors (Olivier 2010). This generated a strong interest in partnership and multi-sectoral collaboration as visible in the development and health literature. Global health and development agencies (such as WHO, GFATM, World Bank, and UNICEF) began to “…turn to religious bodies who might offer what is otherwise lacking� (Cochrane 2006, see also Schmid et al 2008, Woldehanna et al 2005). The WHO stated in its 2004 World Health Report that: “Faith- based organisations have a crucial role to play in the widespread uptake of HIV/AIDS treatments…(they) could be brought into treatment scale-up in order to combine their comparative advantages.� It is from this period onwards that the theme really emerges that the ‘faith sector’ holds untapped resources not otherwise available in other sectors. Alongside this renewed interest in REs from the international development and health communities, came a rapid realisation that due to historical neglect, there was a significant knowledge-gap on religion, the nature of REs, their response and (potential) impact on HIV/AIDS intervention. Indeed, the intersection of religion, health and development has become renowned for its substantial gaps in systematic information and basic data (Schmid et al 2008). However, seemingly disconnected from the struggle to develop better evidence on the role of REs, funding targeted at REs engaged in HIV/AIDS in Africa was increased significantly. As Christoph Benn of the Global Fund to Fight AIDS Tuberculosis and Malaria (GFATM) said in 2003, “for decades they were at best tolerated but not actively supported. Now there is almost a competition among big secular donors to fund the best programmes.� Another example is the US President’s Emergency Plan for AIDS Relief (PEPFAR) launched in 2002, which profiled religious organisations, acknowledged the concept of ‘spiritual care’ and entailed particular funding in relation to religious activities (PEPFAR 2009, Formicola et al 2003). Many of the other large multinational agencies similarly took the decision to target funds to religious organisations - described as a reliable and efficient means to impact on health crises and HIV/AIDS (Taylor 2005a, 2005b, 2007). Increased funding for FBOs came several steps ahead of attempts to fill the knowledge gap, and actions were taken despite little supporting evidence for targeted engagement (Olivier et al 2006). Liebowitz (2004) concurs: “…increasing resources are being devoted to supporting FBOs in global campaigns…Yet remarkably little guidance is available for policy makers on exactly what the strengths of FBOs are, what best practices have allowed specific FBOs to achieve significant successes, and how FBOs can be integrated into broader campaigns for prevention and mitigation…� Said 30 differently, funding and action are mainly based on perceptions (such as those outlined above), and the individual experience of decision-makers, rather than on a body of evidence detailing the comparative value of REs. This is not to denigrate such decisions, but rather to point to lagging research and scholarship that could not at the time provide the underlying evidence to the scale and detail necessary for policy-level decision- making. The latest phase: Mixed opinions and calls for evidence The acknowledgement of this knowledge gap (as well as broader effects of the HIV/AIDS research industry) has led to renewed inquiry at the intersection of religion, development and health, and has resulted in the emergence of a new body of literature addressing the religious response to HIV/AIDS in Africa. An analysis of this literature shows an array of directly conflicting conclusions about REs (Olivier 2010). Or as Dilger (2009) puts it, “in recent years, relationships among religion, development, and globalisation have been discussed critically with regard to the potentially beneficial as well as detrimental opportunities that the work of faith-based organisations (FBOs) presents in relation to HIV/AIDS.� However, a strong discursive device that can be seen in this literature is for the comparative strengths and weaknesses of REs engaged in HIV/AIDS to be listed, frequently in the opening paragraphs, but also usually unreferenced, and commonly prefaced by phrases such as ‘it is accepted that’. Commonly listed strengths might be that REs have extensive infrastructure; reach and access; that they provide services in inaccessible areas; that they have access to dedicated volunteers and educated leadership; that they have unique credibility, trust and acceptance in communities; that they have well-developed networks extending from international to grassroots communities; that they provide a special kind of care; or that they have particular resilience and durability. For example Byamugisha (in WHO 2004) says: “We have a unique presence and reach within communities. We have unique structures and programmes that are already in place. We are available. We are reliable. And we are sustainable. We were there long before AIDS came and we will still be there when AIDS goes away.� In contrast are similar (but shorter) lists of potential weaknesses or liabilities, which make them comparatively poor partners for HIV/AIDS intervention. The weaknesses most commonly listed are limited resources, limited capacity, limited skills (in particular poor documentation), and such concerns as proselytizing tendencies. Again, such statements are usually unreferenced, difficult to track, and rarely tied to any apparent systematic evidence base. Furthermore, these conflicting statements about the comparative advantage or disadvantage of engaging with REs in HIV/AIDS response are likely to appear in the same text. The real challenge is that based on this literature, any position or conclusion about REs and the religious response to HIV/AIDS can be defended. As Olivier (2010) notes, based on this body of recent literature, it is possible to substantiate almost any broad generalization about the religious response to HIV/AIDS. Given the lack of a systematic evidence base, generalisations and stereotyped representations are given more power. In her unpacking of what she calls an ‘epidemic of signification’, Treichler (1992) points out that while HIV/AIDS is a complex cultural phenomenon that produces diversity and 31 contradiction, dominant meanings also emerge - default meanings that can be expressed with little fear of being challenged. In the literature addressing the religious response to HIV/AIDS, the dominant meanings around the comparative advantage tend to be in direct opposition with equally powerful dominant meanings about comparative disadvantage, and researchers and policy-makers are left to negotiate these opposing views. After conducting a review of the literature on religion and HIV/AIDS, Haddad et al (2008) could only conclude that: “there has been a recent boom of interest in the potential of religious entities in establishing effective HIV and AIDS interventions. This interest usually reflects a strongly positive attitude towards working with religious entities and simultaneously some cautionary note, based on perceptions of the potential negative effects of religious messages� (in Taylor 2007). This is typical: data are uneven, diverse, and greatly lacking in specificity –rarely defined enough to know which kinds of REs are more or less suited to which kinds of HIV/AIDs programming. The chronological way these three ‘stages’ of discourse towards the comparative value of the religious response to HIV/AIDS in Africa have been presented is possibly misleading - given that many of these themes linger today. However, it does appear that there has been some shift from early in the HIV/AIDS pandemic, when there appeared to be little targeted interest in how REs might add comparative valued and a generally negative perception of REs; to a more enthusiastic discourse about the positive value REs; and finally to a conflicting body of uneven evidence that provides strong but equally conflicting assumptions about such comparative value. Most recently, there have been signs of a new phase emerging in the international policy and research environment. At many of the recent meetings of international actors on religion and development, there was strong mention made of the need for ‘real evidence’ of the comparative value of religion and REs (CIFA 2010, TBFF 2009, WHO-CIFA 2009). The World Bank, for example, has undergone a transition in its relationship with REs, from a focus on dialogue with faith leaders to more policy-relevant empirical work with REs in the context of country-specific work. Statements have been made in international meetings alluding to a loss of patience for dialogue about the ‘peculiarities of FBOs’ and increasingly asking REs to show empirical evidence of their presence and the comparative value of their work. As noted in one recent meeting in Geneva, “More effective and compelling mapping, including data on value-added by religious health assets, are needed to substantiate basic health services and demonstrate advantages and quality of care among all providers…How can we expect donors and governments to invest in ‘unknown’ or spurious services based on anecdotal evidence alone?� (WHO- CIFA 2009). Tensions that remain It is ironic (or perhaps just a reflection of increasing maturity) that despite a greatly increased body of literature on the religious response to HIV/AIDS in Africa, it appears to be increasingly difficult to make any generalized statements about the comparative value of REs, or specific conclusions that could be implemented at the policy level. It is clear that there is still a critical lack of systematic evidence on the presence, role and impact of religion and REs. Comparative and contextualized evaluations of specific 32 initiatives and institutions are urgently needed, as is the continued collection of base-line evidence through broad-scale tools such as national surveys. One of the areas that remains most critically under-researched is that which could speak to the presence or impact of religious initiatives at a community level. Said differently, while the more ‘formal’ faith-inspired health and development institutions and organizations are slowly being recognized in a more comprehensive way, the community level is where less formal initiatives emerge and the impact on local populations remains critically under-researched, as noted by Schmid et al (2008) in their landscaping study of religious health assets in Africa. It is actually a paradox that it is on this precise level of the local community religious initiatives that we know the least, while it is also (it can be argued) here that the real comparative value of the ‘faith sector’ likely must lie – such as local trust, access, reach and acceptance, potentially impacting on both service provision and behavior change strategies (Olivier and Clifford 2011). This paradox also reminds us that it is at the community level, that some of the most contentious issues for possibly religionophobic policy-makers are most visible. At the community level you are less likely to be engaging through international religious institutions that can appear almost identical to other non-governmental organizations, but rather through local religious leaders, local religious communities, or through ‘faith- forming entities’ such as congregations, churches and mosques – all with unapologetic religious characters, values and priorities. Another reason for a continued lack of impact at a policy level, despite an increased literature – is that the collaborative arena in which this evidence gathering is being conducted is a highly charged and complex context. The issue of proving the comparative value of religion and REs in the response to HIV/AIDS has become a highly visible topic with different agendas influencing what gets researched and by whom. The idea of REs having a comparative value (which is the basis for fundamental issues such as calls for increased funding) is linked to both powerful advocacy groups pushing for the increased recognition of the previously invisible ‘faith sector’, and also to possibly religionophobic agendas. The issue of just what ‘real evidence’ is in this context, or who should be gathering it, is therefore not without contestation. For example, there is a trend in the literature to speak of ‘secular analysis’ – such as Luker (2004) saying, “although (churches) are major institutions…very little secular analysis of their contemporary social capacities and roles is available.� Woldehanna et al (2005) also express some of the tensions, saying, “…The increasing involvement of FBOs in delivering HIV/AIDS services and prevention activities warrants a balanced and impartial examination of their contributions to help optimise their future involvement and collaboration.� The ideas expressed of ‘secular’ and ‘impartial’ research shows some of the tensions at play, and how the collection and analysis of evidence in this field of enquiry cannot be considered to be value-free (this not being about the methods of research, but about who is engaged in this work, and how findings are taken further). Many REs also have historical motivations for being protective of information about their activities and resources. The comparative 33 disadvantage often expressed – that REs ‘are poor at documentation’ – might, for some, be more about a historical lack of trust or alignment with national evaluation and informational systems (see Haddad et al 2008). While there have been several calls for more ‘mapping’ research of REs, it should not be assumed that all REs would automatically see the benefits of being included on the international policy ‘map’ (Olivier 2010). Representation continues to be a critical issue, with REs expressing concern that the current system of networks which are supposed to represent the ‘faith sector’ at a national and international level are over-representative of certain formal and denominational groups, and under-representative of other religious groups, including (for example) the burgeoning Pentecostal groups in Africa who do not usually have clear denominational infrastructures or representation (Haddad et al 2008). Advocating generally for the increased recognition of the presence and impact of the ‘faith sector’ can also have unintended consequences in this complex collaborative environment. For example, in the literature addressing the religious response to HIV/AIDS in Africa, as well as in the broader literature on religion and development, there is a strong theme which names RE as an ‘untapped resource’ that needs to be better utilized or leveraged. Such references have been made by many multilateral development agencies, including the World Bank. Literature emerging from religious groups similarly speaks of their ‘untapped resources’. One unintended consequence of such themes is that many religious leaders are increasingly expressing concerns that they feel ‘used’ by the instrumental way in which they are being drawn into government HIV/AIDS programs (Haddad et al 2008, Olivier 2010), or being utilized for a particular set of assets such as volunteers or specific programming, without concurrent recognition of the particular values or purpose that drives their work. Again, there is tension between those advocating for the faith sector to be more fully recognized, often using the unacknowledged provision of health services as a motivation - and the unintended result of an international community seeking to leverage these resources in secular-oriented policy environments. Perhaps one of the reasons the growing body of literature on the religious response to HIV/AIDS in Africa has not resulted in significant policy-level action, is because of such tensions which hamper the gathering and utilization of evidence. Such issues cannot be overcome through simple strategies such as increased volume of research or increased M&E training for REs (only) - but is rather about collaborative cooperation around evidence gathering, trust, and how REs see themselves represented in the international information and policy arena. Said differently, the development of a systematic evidence- base on the religious response to HIV/AIDS is not possible without also negotiating this complex (historical) collaborative environment. Gaining access to baseline data as well as the actual utilization of completed research should not be taken for granted. However while such tensions and collaborative concerns can be negotiated – there is also another fundamental issue which inhibits policy engagement, that of clashing typologies and frameworks being applied to this evidential field. It is to this issue that we turn next. 34 EXISTING TYPOLOGIES AND THEIR LIMITS Language and terms: confounding religious categories Assessing community responses to health and HIV/AIDS is already complicated because communities themselves are complex and assessments are necessarily context-specific. In fact, it is not always useful to suggest what a ‘typical’ community response is or should be. As Rodriguez-Garcia et al (2011) put it: “The richness of the community response may very well be in its multiple combinations and its variety - its uniqueness in the community’s cultural and geographic context.� However, these authors continue by noting that at the same time, for policy-makers, some systematization and simplification is necessary in order to suggest broad tendencies, and come up with diagnostics of the strength and potential weaknesses of existing responses. It is also acknowledged that religion and the myriad ways in which religion manifests itself in local communities often require qualitative and textured forms of description and research. However, in the same way, it is still necessary to also strive towards some level of systematization and quantification if religion and REs are to be acted on at a policy level. The assessment and gathering of systematic evidence of religious community response to HIV/AIDS may be particularly hazardous, given the historic neglect of the research field followed by a rapid resurgence of interest. As a result of this interest a variety of lenses have been rapidly applied to religious communities and institutions, with rarely any two studies using the same typology, classification system or unit of analysis - and many taking transverse slices across traditionally separate fields of inquiry (for example, by investigating the ‘faith sector’ including all entities from congregations to mission hospitals, with no equivalent comparative made of these combinations in the ‘secular sector’). With no clearly defined field, diverse lenses and units of measurement have been applied to the ‘religious response’ - resulting in diverse and uneven evidence. The intersection of religion, health and development has long been hampered by differences in language (of discipline and culture rather than linguistic). As Benn notes (2009), “Sometimes the biggest challenge for forging collaboration around this theme between public health experts, representatives of international organizations, and the faith community is the lack of a common language and terminology, which often leads to misunderstandings and frustrations.� The most obvious example of this communication gap is the exceptional lack of clarity and consensus on the basic ‘unit’ – say, the ‘faith- based organization’ – to be assessed. Scholars have argued for some time that there is simply no such thing as a ‘typical NGO’ (Martens 2002) – and a lively discussion has ensued around classification systems. However, this is completely eclipsed by the vigour of the debate over what a ‘faith-based organisation’ should most appropriately be named – which continues to rage on both scholarly and the international policy fronts, seemingly without any consensus. Beyond the academic realm, global health and development institutions, governments and REs themselves have all weighed in through attempts to reshape and redefine what REs are, and what they should most appropriately be compared with. A recent example of lack of consensus can be seen in UNAIDS’ Strategic Framework for Partnership with Faith- based Organizations (UNAIDS 2009), which reports on the results of an 18-month 35 consultation process during which terminology was discussed extensively with UNAIDS partners. Although the document does describe a particular terminological framework, it then concludes that: “Each UNAIDS Cosponsor may have reasons for its own specific terminology and engagement. For ease of reference in this framework, the term ‘faith- based organization’ will generally cover the various categories listed above, except where indicated otherwise.� This demonstrates both the continued lack of consensus among collaborative partners – as well as the difficulties of finding appropriate nomenclature. This is not only a debate about terms, but also about substance. The scholarly debate over the ‘FBO’ (Bradley 2009, Berger 2003, Clarke 2006, Sider and Unruh 2004), appears to have had little impact on the parallel policy or operational environments, where few meetings are held where the terminological issue is not (re)addressed (Olivier 2011). For example, a typical statement showing some of the concerns can be seen in a symposium report from the Center for International and Regional Studies (2008): “…the group focused initially on exploring the significance that should be given to the terms and concepts of ‘faith’, ‘faith-based’, and ‘faith-inspired’, and the significance of describing organisations or communities as Muslim or Islamic, or non-denominational or secular…[and was admonished] to pay special attention to vocabulary and especially terms that may be imbued with western framing and historical legacies…The crux of the issue lies less in how an individual or an organisation defines their ‘faith’ motivations than on how others interpret and assess its significance. The topic is strewn with pitfalls, and virtually all terms and categories are slippery and problematic.� One of the main challenges is that terms such as ‘FBO’ or similar terms 8 are now being used interchangeably to indicate any range of meanings – indeed, any entity with a partially religious character. For example, showing some of the range, a World Council of Churches document (Lux and Greenaway 2006) states: “The term FBO is used here to describe a broad range of organisations influenced by faith. FBOs include: religious and religion-based organisations and networks; communities belonging to places of religious worship; specialised religious institutions and religious social service agencies; and registered and unregistered non-profit institutions that have a religious character or mission. They might be small, grassroots organisations with simple structures and limited personnel or large, global institutions with highly sophisticated bureaucracies, wide networks, substantial financial resources, and significant human capacity. In some cases they are led by clergy…in other cases laypersons…� Terms such as ‘FBO’ can be used broadly in this way, and also specifically for a range of different sub-categorizations. A landscaping literature review of religious health provision in sub-Saharan Africa (Schmid et al 2008) noted over three hundred terms 8 There are currently several terms used to broadly describe an entity with a religious or faith dimension, and that distinguishes them from secular entities. The most commonly used term is ‘faith-based organization’ (FBO), but alternatives include ‘faith-inspired organization’ (FIO), and ‘faith-based initiative’ (FBI), as well as many other terms (see Olivier 2011a). There are also a plethora of abbreviations for secular or broader categories, including civil society organization (CSO), community-based organization (CBO), non-governmental organization (NGO), religious non- governmental organization (RNGO), and even community support group (CSG). 36 being used to describe REs engaged in health care in Africa. A review of the literature addressing the religious response to HIV/AIDS in Africa indicates similar variety – and six common meanings given to ‘FBO’ (none discrete): 1) faith-forming entities (whose primary function is the formation of faith or worship), 2) religious leaders, 3) Religious nongovernmental organizations, 4) community-based religious initiatives, 5) networks, and 6) health facilities. At this time, there is no consensus on the terminology and definition of REs involved in development and HIV/AIDS, in Africa or internationally – and the trend has arisen where authors redefine the terms in each individual piece of literature; utilize different terms in different outputs from the same empirical research; or even change terms mid-document without explanation (Olivier 2011). Mapping work on the religious response to HIV and AIDS at a community level has forced researchers to confront these inconsistencies and the inadequacies of our current frames of assessment. In the rest of this section, we illustrate just a few of the main confounding difficulties with some of the most common classification approaches. 9 Classification by form One challenge for typologies of the response to HIV/AIDS is the identification of ‘form’. In many development contexts, at a local level, REs are rarely constituted in the form suggested by terms such as ‘institution’ or ‘organization’ and just as frequently appear as fluid formations of individuals, initiatives or responses (ARHAP 2006, Thomas et al 2006). Even the classification of entities whose primary purpose is the formation of faith is problematic: since even terms such as ‘church’ or ‘congregation’ are not without difficulties, given that the formality suggested by such terms is undermined when faced with informal religious ‘congregations’. The concept of ‘congregation’ is indeed much more fluid in multi-religious settings than in most Western contexts. In African settings, for example, communities may simultaneously negotiate and utilize different religious community resources – as evidenced by plural health-seeking behaviors where individuals commonly utilize multiple ‘healing’ strategies such as faith-healing and traditional-healing alongside biomedical care in response to HIV/AIDS (ARHAP 2006). There is also some tension with REs being ascribed to different sectors. As Melkote and Steeves (2001) say, the objects of development are often “inserted into implicit (and explicit) typologies which define a-priori what they are…Third World countries became pliable objects to be manipulated by the development experts.� This is particularly apparent in the HIV/AIDS context in Africa – where some REs are defined as part of the development response (as religious NGOs, RNGOs, part of the civil society sector, collaborating and being supported through Multisectoral AIDS Councils); and others are defined as part of the medical response (as health facilities, part of non-state service provision, collaborating and being supported through Ministries of Health) (Haddad et al 2008). What is challenging is that there is only a limited logic to these classifications, and 9 See also Rodriguez-Garcia et al (2011) who use the following criteria for identifying community-based HIV/AIDS responses: (1) the types of organizations and structures implementing the response, (2) the types of activities or services implemented and the beneficiaries of these, (3) the actors involved in and driving community responses, (4) the contextual factors that influence community responses, (5) the extent of community involvement in the response, and (6) the extent to which community responses involve wider partnerships and collaboration. 37 more confounding outliers than neatly compliant entities. For example, Christian Health Association members typically get named and counted both as part of civil society response to HIV/AIDS (RNGOs engaged in development) and part of the health system providing a medical response to HIV/AIDS (ARHAP 2006, Schmid et al 2008). Being classified as a health facility or a RNGO has consequences for critical issues such as representation and access to government and external resources. Mapping studies have uncovered the presence of a significant number of REs working against HIV/AIDS not previously recognized by national and international health institutions – in particular a mass of smaller community-based religious initiatives often not recognizable as NGOs or health-facilities, more difficult to measure and understand, and varying considerably in different contexts (ARHAP 2006, Haddad et al 2008, Schmid et al 2008). This is a significant hidden group of initiatives, differently described as informal, unorganized, non-mainstream, non-facility-based, or community-based (Foster 2004, Schmid et al 2008). One reason for their continued invisibility could be because they fit less comfortably with current nomenclature and classification systems. For example, a women’s group working out of a congregation without the knowledge of their religious leader, or a neighborhood initiative using bicycles to transport AIDS patients to the local hospital each week for treatment – are less easily mapped than CBOs, NGOs and health facilities. We know significantly more about the forms of religious community initiatives which we are comfortable in seeing – such as RNGOs or health facilities with a recognizable form and facility. These are the initiatives which have so far received the greatest research focus, and which clearly receive the most support, which in turn results in the establishment of more initiatives of these kinds (for example, a boom of RNGOs has been noted since around 2000 in many African countries such as Zambia, Malawi, Kenya or Uganda – ARHAP 2006, Haddad et al 2008). There remain huge challenges to our research strategies and tools to be able to adequately map and support informal initiatives without (even inadvertently) destroying them or reshaping them by imposing rigid frameworks on them. Classification by function Attempts have also been made to classify REs by their geographic location, size, scope, reach or primary activity (Berger 2003). There are of course contextual similarities and patterns to be found. For example, there are hints at regional patterns that differ between Anglophone, Francophone and Lusophone countries. Many francophone West African countries demonstrate a particular range of civil society institutions and initiatives which includes fewer religious health facilities or RNGOs, and community level response more frequently channeled through religious congregations (such as mosques) and religious schools. In comparison, Southern African countries with similar colonial and historical contexts, such as Zambia, Malawi or Kenya demonstrate an abundance of RNGO and health facility-type initiatives (Schmid et al 2008). Still, classifications by function show strain when faced with the reality of complex community contexts. For example, regional categorizations are problematized by the diverse transnational and transregional ties among REs with their unique connections for collaboration, resources, funding or ownership (Schmid et al 2008). REs often work 38 across national boundaries, collaborating in complex regional and transnational networks that fit poorly into national scale assessments (Haddad et al 2008). Representation (e.g. local, regional, international) is another category that does not fit comfortably - as there are many REs that are simultaneously represented as part of a local program, a regional denomination, and an international network. Or vice versa, international REs that simultaneously act in a local community as an intermediary (channeling funding to other local initiatives) and operate their own local programs (therefore sometimes perceived as competition by the same local initiatives) – making it difficult to map these initiatives in any simplistic way (ARHAP 2006). Mapping studies have also suggested that religious initiatives are often fluid in nature, adapting to the needs around them and shifting focus, making it particularly difficult to categorize their type or activities over time (Olivier et al 2006, Schmid et al 2008). That is, while many religious community initiatives are shown to have been in place for a long time, they are also often extremely adaptive to need and available resources – changing character faster than they can be mapped (ARHAP 2006). Many, if not most, HIV/AIDS- engaged REs in sub-Saharan Africa have demonstrated broad, holistic portfolios that simultaneously range across a number of primary activities; some such activities might very well be religious, while others are more recognizable with a development or public health lens. For example, there are Muslim communities financially supporting a government hospital; or traditional healers (who might also be Christian pastors) running ART and referring patients to government hospitals (ARHAP 2006, Schmid et al 2008). Religious communities also tend to have a particular language in which they describe and evaluate their care and support activities, which often might extend beyond what is typically defined as ‘care’ – in particular the framing of ‘spiritual care’ as a core part of care and support (Olivier and Clifford 2010). Classification by ‘religiosity’ Perhaps the most problematic classification approach is the basic binary division between religious and secular – which is especially troublesome given that the terminology (and literature) described above rests on the assumption that religious initiatives can (and should) be separated from their secular comparatives. There are several available models which classify an organization by its level of ‘religiosity’. For example, the World Council of Churches has differentiated between ‘faith-related organizations’, ‘faith- background organizations’, ‘faith-centered organizations’ and ‘faith-saturated organizations’ (Doupe 2005, see also Sider and Unrah 2004). Some studies have sought to show how religious organizations can be classified on such a scale, or how the activities in which an organization engages may affect how it fares on such as scale. For example, work in the US context has suggested that HIV/AIDS-engaged organizations might lose their religiosity as they become more dependent on ‘secular’ funding (Chambre 2001). However, while measures of ‘religiosity’ may appear relevant in communities with more fixed congregational structures, such as American congregational studies, they become more problematic in development settings, and when engaging with local communities (Olivier 2010). 39 The underlying problem, is that the Cartesian division between ‘religious’ and ‘secular’ cannot be assumed in many African contexts where religion is embedded in everyday life, and also integral to the character of many secular-classified organizations. An institution which a researcher or policy-maker might assume to be secular may insist that they are religious. Just as an illustration, at a workshop on mapping religious health assets in Kampala, a member of the Infectious Disease Institute (attached to the university) insisted that “…we are a faith-based organization…all organizations in Uganda are faith-based� (ARHAP 2007). Conversely, the Aga Khan Development Network, a large and power Islamic hospital system and development agency, states emphatically that it is not a religious institution. There are also community level initiatives whose religious identity is undefined – that is, describing themselves as deeply religious, but not affiliated to any particular religious group - for example community support groups that are providing HIV/AIDS care in Lesotho but are not tied to any structure or organization and receive little or no external support (ARHAP 2006, Liebowitz 2002, Schmid et al 2008). Some studies have attempted to assess FBOs by ownership, for example by which denomination, faith tradition, or coordinating network the entity might belong to. This is also hazardous. For example, in many countries, faith-inspired health facilities have been designated as district hospitals (that is, as part of the public system), and vice versa, for example, Kilembe Mines Hospital in Tanzania is owned by a parastatal body but is managed by the Catholic Diocese of Kasese (Schmid et al 2008). It is in fact common for health programs and facilities to be owned or operated by more than one network or body. While it might seem easy to identify a local ‘congregation’ – very often it is the offshoot activities that are of most interest (especially in relation to measuring HIV/AIDS response), and mapping studies have shown that it cannot be assumed that the religious leader is aware of such off-shoot initiatives (and are possibly more likely to be involved if there is funding - see Agadjanian and Sen 2007). Classification is nowhere a value-free activity, and institutions and initiatives often adjust descriptions of their type and activities according to how they perceive it would be most useful to be understood in a particular situation or context. Classification has clear implications such as access to resources, representation and collaboration (for example with government). At one time it might be useful to be labeled an ‘FBO’ to open certain funding doors (such as PEPFAR), and at other times less so, for fear of “potentially negative connotations associated with religious references as well as legal obstacles that arise when applying for public funding� (Berger 2003). We do not mean here that the religious-secular classification should be ignored or that we should return to a state of categorical religion-blindness (even were that possible). However, one of the implications of these examples emerging from community mapping studies is a warning that in the process of gathering systematic evidence of this previously neglected field, we might have become too caught up dialoguing about the comparative value of ‘FBOs’, and have invested less energy in gathering evidence on how religion and religious values might work as a motivational force or have an impact on behavior in and through institutions and initiatives not currently classified as ‘religious’. In a research study in Zambia and Lesotho, local communities identified the 40 exemplar community initiatives working in HIV/AIDS as those who both demonstrated (the expected) solid programmatic, operational, and associative characteristics; and then infused these with ‘intangible’ religious factors such as a particular kind of compassionate care or hope (ARHAP 2006). This implies that to understand local community initiatives in Africa, it is often necessary to go beyond standard evaluation measures when we compare REs to their ‘secular’ counterparts, to also consider creative ways for evaluating and better understanding ‘the religious’ elements and values in initiatives across the board. As ARHAP (2006) argued, “…engagement with REs must not be with an eye to ‘convert’ them to exemplary NGOs and diminish the very intangible dimension that distinguishes their contribution to health…The opportunity is not to turn public structures into religious ones or to turn religious structures into public ones, but to gain the benefits of the alignment of their respective strengths for the community.� 10 Finally, REs do not function in a secular vacuum, and it is important that religious response be evaluated as part-of local community, rather than apart-from. This is emphasized by a local mapping exercise in Zambia and Lesotho, where in every site, local health seekers emphasized how religious initiatives always operate in partnership with a range of ‘partner initiatives and institutions’ (ARHAP 2006). The real challenge is therefore how to gather systematic evidence that is inclusive of religion and religious initiatives, but in doing so does not cut these off from the comparative context of the communities in which they are located. Summing up: Generalization is dangerous All this talk of typologies and classifications may seem a bit too much like a descriptive word game – particularly frustrating for those dealing with the everyday challenges of working in resource-constrained environments. However, these ‘words’ are the building blocks upon which international and national policy relevant dialogue are built. The battles over naming REs appear to be less about finding the best terminology, and more as a result of ill-fitting and chafing evidential frameworks – highlighting the fragility of our knowledge systems and methods of inquiry. We provide here no ready answers - and certainly no suggestions for new terminologies - but it is worth noting that the continued debates over finding a broad generic term for REs or a sustained set of subcategories has been unhelpful and may have exacerbated the lack of systematic evidence and policy engagement. This is less a concern about reaching a mass consensus on terminology, and more about serious assessment of the basic units of analysis and the frameworks being applied. Only then can a more systematic (and comparative) evidence-base be established, for more relevant and strategic policy engagement. 10 The African Religious Health Assets Programme (ARHAP) has suggested the ‘religious health asset’ (RHA) framework and language, which seeks to circumvent some of these classification issues. An RHA is described as a (tangible or intangible) asset located in or held by a religious entity that can be leveraged for the purposes of development or public health. The approach goes beyond the more easily identifiable tangible assets, e.g. buildings and equipment, to include what have been described as ‘intangible’ health determinants. These include aspects of health behaviour modification programmes and the beliefs, values and commitment of health workers. The focus on intangible assets makes possible an assessment of the contribution of FBOs not merely as alternative health service providers (see Schmid et al 2008). 41 As a result of the transverse slicing of the field, where an interest in the ‘faith sector’ is not rooted in any single field, interest or particular set of indicators – terms such as ‘FBO’ have generally lost any specific meaning or definition, and this makes any comparison of data between studies of the religious response to HIV/AIDS in Africa difficult, if not impossible (Olivier 2010 and 2011, Schmid et al 2008). As a result, there is also little consensus on which parts of the ‘faith sector’ are best targeted for specific strategies, which areas require more support, or where comparative strengths and weaknesses might be evidenced. It should be clear by now that generalizations about the generic ‘faith sector’ or the generic ‘FBO’ might be a powerful advocacy tool, but has limited value as an evidential or analytical measure. How necessary is it really to prove that all REs are comparatively more networked? Or that all REs are weak in the area of documentation? The huge variety and contextual variation within the ‘faith sector’ makes any generalized statement about the comparative value of REs immediately suspect and open to critique. TOWARDS OPERATIONAL TYPOLOGIES: AN ILLUSTRATION It should be clear from the two previous sections that the assessment of the religious community response to HIV/AIDS appears through a diverse and at times opposing array of lenses. This lack of consensus is probably one of the reasons why existing frameworks and typologies have had so far a limited ability to inform policy-level engagement, especially that relating to operational action. They also have not helped as much as was hoped in providing a more systematic understanding of the response itself. Given that the HIV/AIDS thematic is an area around which some policy actions still remain controversial, at least for some, disagreements in perceptions and points of view are likely to remain for some time. But this does not mean that progress cannot be accomplished. Consider the case of methodological approaches. There is still tension in getting the right balance between qualitative, context-rich information which can describe different formations of community response, and systematic quantitative assessments which require standardization of data and language. But there seems to be more consensus today that both forms of empirical evidence are needed – especially in evaluation of HIV/AIDS response. What is then missing is the toolset that could more adequately integrate multiple (and diverse) data streams or dimensions for systematic analysis of complex community contexts. Such integrative methods are required to answer specific questions such as which population is best served by religious initiatives, or what impact individual religious leaders may be having on behaviour change for HIV/AIDS response. Acknowledging the challenges and inherent complexity of religious community responses, the gathering of baseline data focusing on integrating data streams and frameworks could help move the policy discussion further. In this last section, our objective is very limited. We hope to suggest with a simple conceptual illustration how the combination and visualization of various data sources or typologies could indeed be useful in providing more operationally-oriented typologies. It is of course much easier to review and point to the limits of existing typologies, as we did 42 in the previous section, than to provide new ideas that help overcome some of those limits. We are conscious of the fact that what we are suggesting in this section remains embryonic, but the idea is to try to show how some of what has been discussed in the literature can be harnessed in more user friendly ways so that the information can be actually used by policy makers faced with difficult trade-offs. Essentially, what we suggest is to combine various dimensions or typologies in such a way that they may highlight what exists on the ground, and what is missing to help households cope with HIV/AIDS. This is done using matrix-based visualizations, as is often done in management practice. We provide only one example here, but many alternatives could be proposed as well. In the review of the literature in the previous section, two of the distinctions that were made when describing the role of REs were first the classification by form, pointing especially to the range existing REs from purely informal to purely formal organizations (i.e., who is providing services and benefits), and second the classification by function, which relates to the types of services that are being provided (i.e., what services or benefits are being provided in various areas, including prevention and care, and by whom). Both classifications are important for policy. Ultimately, policy focuses on the types of services to be provided. In addition, because policy requires institutional instruments and accountability frameworks for implementation, forms – whether REs and other providers are formal or informal - also matters. But what we would like to suggest is that the combination of these two types of classifications may provide more insights for policy makers than each of the two classifications taken on its own. One more important methodological point must be made regarding the scope and nature of the application of typologies aiming to inform policy. We are considering the context of a secular state that does not have any aims pertaining to religion per se. This means that any new typology or combination of new typologies could consider the ‘comparative value’ of REs, but only in its secular aspects. For example, it could be that faith itself is the key behind the success of a particular intervention, and that this intervention in turn strengthens the faith of those exposed to it – which is perhaps in large part why REs implemented the intervention in the first place. But a secular state would probably (in most cases) fund only the ‘secular’ component of the intervention, and judge its outcome only in secular terms. 11 In addition, a secular government aiming to support and coordinate the overall response to HIV/AIDS should clearly map all of the activities that may contribute to specific objectives in this area, and not only those managed by REs. That is, in seeking a comprehensive and systematic understanding of community responses to HIV/AIDS, it is necessary to be inclusive of a wide range of forms of community initiatives in order to account for both formal organizational response as well as informal community initiatives which might not have any organizational (or physical) structure and be significantly more fluid in nature, and it is also necessary to include both secular organizations and REs, as well as whomever falls in between (under the religiosity typology). 11 This is admittedly quickly said, and the issue of what a secular state should and should not do in interacting with religion is much more complex, but we wish not to delve in this here. 43 Consider then as an illustration of policy-oriented typology the matrix in Figure 1, which combines the classifications by form and by function. Community initiatives – whether religious or not – are placed on the vertical axis on the basis of their form: whether they are formal, informal, or a combination of both or some intermediary nature. We say then an intervention is formal if it is implemented by an organization with some form of institutional infrastructure and status. This might be a facility, legal registration, bank account, management committee, defined responsibilities, paid staff, strategic plans, or benefit from external support (e.g. CBOs, NGOs, or health facilities). We define as informal the myriad varieties of initiatives and responses that usually do not conform to these criteria and thus typically would not have a facility, organizational infrastructure, or even necessarily a name, often relying on voluntary effort, and operating without external support. By emphasizing the level of formalization we circumvent some of the problematic labels, and rather focus evidence gathering on what capacity and resources are available, and in what form. This does not imply that one form has greater value than another, but rather that for strategic purposes, it is critical that policy makers know, for example, whether the interventions are being enacted by a faith-inspired university hospital, or an initiative being run out of a local congregation. As many studies have shown, the local and community informal responses are frequently more powerful at a local level. But lack of formality also often makes it more difficult for Ministries or local authorities to support such initiatives. One of the lingering questions, therefore, is how to ‘formalize’ (at least partially) such initiatives without destroying the very fabric of the communities that helped them flourish, or the flexible and supportive nature of their initiative. A second critical differentiation, especially in the area of HIV/AIDS, relates to the kinds of activities in which REs and other organizations are primarily engaged, or the function they serve. As argued above, the literature addressing the religious response to HIV/AIDS demonstrates two main concerns for the comparative value of religious community initiatives – on the one hand, service provision, and on the other, the potential for faith communities to influence individual or community behaviors and attitudes, particularly in relationship to HIV/AIDS. It should thus be helpful for those working at an implementation or policy level to know which REs and other initiatives are more geared towards service provision (such as a home-based care group, or the extensive non- state service provision provided by the Christian Health Associations), and which have more potential in areas of community engagement and impact on behavior and attitudes (such as an individual religious leader or a small group of women from a congregation influencing sexual behavior). Again, this can be expressed as a range – given that many programs are engaged in both types of work, and again, no value judgment should be made as to whether service delivery is more important than work related to behaviors and attitudes. Different community activities (religious or not) could be mapped onto such a matrix. Each activity would be represented by a circle whose size could be proportional, say, to the number of people reached, or depending on the use of the typology, to the cost of the activity or even the likely impact of an activity on specific targets related to HIV/AIDS if 44 that information is available. Different colors could be used to represent activities benefiting from government support, as compared to activities that do not benefit from such support, or urban-rural service allocations, or a distinction between faith-inspired and secular interventions. Based on the qualitative research done to-date, we would expect that in some community contexts, the matrix might look very different from what is going on in another community, and this can be helpful in assessing which types of policy/program interventions need to be supported more, or created. Figure 1: Illustrative Typology of Community Responses to HIV/AIDS Source: Authors. Note: The matrices illustrate conceptually how two different community contexts might look, and are not representative of any particular country or local context. For example, one could expect that in many Francophone Africa, faith-inspired networks of formal service providers (such as the Christian Health Associations) tend to be less developed and have smaller memberships, while in Anglophone Africa, these networks often account for 30 percent to 40 percent of hospital facilities. We would also expect more formal NGOs to be present at the local level in a typical Anglophone African country than in a Francophone one. But there may be fewer differences in the role of informal networks and interventions between the two types of countries (we simply do not yet know). But the more general conclusion is that this type of visualization may help in assessing whether more support is needed for service delivery as opposed to interventions aiming to change behaviors and attitudes, and whether the type of support 45 provided needs to be adapted to take into account the more or less formal nature of the organizations or groups implementing the interventions. This visualization is admittedly very simple, but such methods can be used as a starting point, provided the evidence is indeed gathered, which should not be too difficult at least at the local level (national inventories are much more complex to assemble given the need to carry out extensive mapping work first). Such visualizations can help identify weaknesses, strengths, gaps and potential areas of overlaps in various areas of the community response, and more generally in re-orienting resources in order to achieve higher overall impact. Importantly, such visualizations seek to blend several different types of data, and can also be useful in making more apparent the inadequacy of assessments which focus only on particular aspects of the religious and other community responses and are not inclusive of the range of other supportive activities in which the REs (and other initiatives) may be engaged. Further work is of course needed to test empirically whether mixed typologies such as the illustration suggested here indeed work, but this is the type of work that seems required today in order to make analysis of the community response more operationally relevant, while still acknowledging the inherent complexity of this response. CONCLUSION The comparative value of religious entities or responses to HIV/AIDS as they emerge from communities of faith is still difficult to prove based on the current evidence at hand. Broad advocacy statements of comparative advantage (or disadvantage) of the ‘faith sector’ or the generic ‘FBO’ should be avoided where possible. Such broad-scale assumptions may do more damage than good as they tend to get immediate pushback from international and ‘secular’ audiences when they are positive – and conversely, broad statements of comparative disadvantage get immediate pushback from religious groups and institutions who do not feel they are adequately understood or assessed fairly. Instead of such broad statements, focusing on specific and targeted comparative assessments at the local level – that is, close assessment of comparative strengths and weaknesses, can help design better and more targeted interventions. It is however important that in the research agenda on the religious response to HIV/AIDS, we do not shy away from complexity. Religion, religious institutions and community initiatives are by nature diverse and complex. And indeed, ‘simple’ solution to the HIV/AIDS pandemic has been challenged. Peter Piot said as he opened the Mexico AIDS Conference in 2008, “Let’s never forget that the epidemic could still bring us new surprises – as it has done so many times already. If we are to get ahead of this epidemic, it is time to come to terms with complexity…� In some cases, it may even be the complexity (or richness) of a contextually-specific local response that provides added value. The challenge is therefore to work towards multidisciplinary methods of describing and assessing such complexity in a way that is not only conscious of the intricate collaborative environment in which interventions and evidence-gathering take place, but also does not weaken the initiatives we are seeking to support through ill- fitting classification or policy-level paralysis. While improving typologies in this area 46 will remain messy and imperfect work, it appears to be worth the continued effort – and although discussions of evidence, typologies and classifications might seem technical or even abstract, at their heart is a basic understanding that there is potential and opportunity here for positive change. REFERENCES African Health Assets Programme (ARHAP). 2006. Appreciating Assets: The Contribution of Religion to universal access in Africa, Cape Town: African Religious Health Assets Programme. _____. 2007. PIRHANA Workshop Report: Uganda, Kampala. Cape Town: African Religious Health Assets Programme. Agadjanian, V., and S. Sen. 2007. “Promises and Challenges of Faith-Based AIDS Care and Support in Mozambique.� American Journal of Public Health 97 (2):362-366. Asad, T. 2003. Formations of the Secular: Christianity, Islam, Modernity. Stanford: Stanford University Press. Benn, C. 2002. “The Future Role of Church Related Hospitals and Health Services in Developing Countries.� Edited by DIFAEM, Tubingen: German Institute for Medical Mission. _____. 2003. “Why Religious Health Assets Matter.� In ARHAP: Assets and Agency Colloquium, Pietermaritzburg: ARHAP. _____. 2009. “The Continued Paradigm Shift in Global Health and the Role of the Faith Community.� In ARHAP Conference: When Religion and Health Align - Mobilizing Religious Health Assets for Transformation. Cape Town: African Religious Health Assets Programme. Berger, P. L. 1999. “The Desecularization of the World: A Global Overview.� In P. L. Berger, The Desecularization of the World: Resurgent Religion and World Politics. Washington, DC: Ethics and Public Policy Centre. Bradley, T. 2009. “A Call for Clarification and Critical Analysis of the Work of Faith- Based Development Organizations (FBDOs).� Progress in Development Studies 2:101- 14. Brundtland, G. H. 2000. “Massive Effort Advocacy Meeting.� Winterhur, 03 October 2000. 47 Center for Interfaith Action on Global Poverty (CIFA). 2010. Many Faiths, Common Action: Increasing the Impact of the Faith Sector on Health and Development - A Strategic Framework for Action. Washington, DC. Center for International and Regional Studies (CIRS). 2008. Symposium: Global Development and Faith-Inspired Organizations in the Muslim World. Doha: CIRS and Berkley Center for Religion, Peace and World Affairs. Chambré, S. M. 2001. “The Changing Nature of ‘Faith’ in Faith-Based Organisations: Secularisation and Ecumenicism in Four AIDS Organisations in New York City.� Social Service Review September: 435-455. Clarke, G. 2006. “Faith Matters: Faith-Based Organizations, Civil Society and International Development.� Journal of International Development 18: 835-848. Cochrane, J. R. 2008. “Fire from Above, Fire From Below: Health, Justice and the Persistence of the Sacred.� Theoria 116: 67-96. _____. 2006. “Religion, Public Health and a Church for the 21st Century.� International Review of Mission 95, no. 376-377: 59-72. _____. 2006. “Understanding Religious Health Assets for Public Health Systems.� Paper Presented at the DIFAM Consultation on Religion and Health, Tübingen, Germany, February. _____. 2003. “Religion as Social Capital in the Context of Health: Mapping the Field.� In ARHAP: Assets and Agency Colloquium, Pietermaritzburg: African Religious Health Assets Programme. Derrida, J., and G. Vattimo. 1998. Religion: Cultural Memory in the Present. Cambridge: Polity Press. Dilger, H. J. 2009. “Doing Better? Religion, the Virtue-Ethics of Development, and the Fragmentation of Health Politics in Tanzania.� Africa Today 56 (1): 89-108. Doupe, A. 2005. “Partnerships Between Churches and People Living with HIV/AIDS Organizations.� Geneva, Switzerland: World Council of Churches. Ellis, S., and G. TerHaar. 2001. Worlds of Power: Religious Thought and Political Practice in Africa. Johannesburg: Witwatersrand University Press. Farmer, P. 1999. Infections and Inequalities: The Modern Plagues. Berkeley, LA: University of California Press. Formicola, J. R., M. Segers, and P. Weber. 2003. Faith-Based Initiatives and the Bush Administration: The Good, the Bad and the Ugly. Lanham: Rowman and Littlefield. 48 Foster, G. 2004. “Study of the Response by Faith�Based Organizations to Orphans and Vulnerable Children.� New York: World Conference of Religions for Peace and United Nations Children Fund. Fox, D. M. 1992. “The Politics of HIV Infection: 1989-1990 as Years of Change.� In E. Fee and D.M. Fox, eds, AIDS: The Making of a Chronic Disease. Berkeley: University of California Press: 125-143. Garner, R. C. 1999. “Religion in the AIDS Crisis: Irrelevance, Adversary or Ally?� AIDS Analysis Africa 10: 6-7. Green, E. C. 2003. “Faith-Based Organizations: Contributions to HIV Prevention.� Washington DC: US Agency for International Development and The Synergy Project, TvT Associates, Harvard Center for Population and Development Studies. Haddad, B., J. Olivier, and S. De Gruchy. 2008. “The Potential and Perils of Partnership: Christian Religious Entities and Collaborative Stakeholders Responding to HIV and AIDS in Kenya, Malawi and the DRC.� Study Commissioned by Tearfund and UNAIDS. Cape Town: Africa Religious Health Assets Programme. Kim, J. Y, J. V. Millen, A. Irwin, and J. Gershman, eds. 2000. Dying for Growth: Global Inequality and the Health of the Poor. Monroe: Common Courage Press. Liebowitz, J. 2004. “Faith-Based Organizations and HIV/AIDS in Uganda and KwaZulu- Natal.� Health Economics and HIV/AIDS Research Division, University of KwaZulu Natal. _____. 2002. “The Impact of Faith-Based Organizations on HIV/AIDS Prevention and Mitigation in Africa.� Health Economics and HIV/AIDS Research Division, University of KwaZulu Natal. Luker, V. 2004. “Civil Society, Social Capital and the Churches: HIV/AIDS in Papua New Guinea.� In Techreport, State Society and Governance in Melanesia Project (RSPAS), ANU. Lux, S., and K. Greenaway. 2006. “Scaling up Effective Partnerships: A Guide to Working with Faith-Based Organisations in the Response to HIV and AIDS.� Church World Service, Ecumenical Advocacy Alliance, Norwegian Church Aid, UNAIDS, World Conference of Religions for Peace. Marshall, K. 2009. “Confronting Religionophobia.� Washington Post, August 10, 2009. Martens, K. 2002. “Mission impossible? Defining Nongovernmental Organizations.� Voluntas 13(4): 271-285. 49 Melkote, S. R, and H. L. Steeves. 2001. Communication for Development in the Third World: Theory and Practice for Empowerment. London: Sage. Olivier, J. 2011. “An FB-oh?: Mapping the Etymology of the Religious Entity Engaged in Health.� In B. Schmid, Cochrane, J. R. and T. Cutts, eds When Religion and Health Align: Mobilizing Religious Health Assets for Transformation. Pietermaritzburg: Cluster Publications. _____. 2010. “In Search of Common Ground for Interdisciplinary Collaboration and Communication: Mapping the Cultural Politics of Religion and HIV/AIDS in sub- Saharan Africa.� PhD Dissertation. University of Cape Town. Olivier, J., and P. Clifford. 2011. “Religious Community Care and Support in the Context of HIV and AIDS: Outlining the Contours.� In B. Haddad, editor, Religion and HIV/AIDS: Charting the Terrain. Pietermaritzburg: Cluster Publications. Olivier, J., J. R. Cochrane, and B. Schmid. 2006. “ARHAP Literature Review: Working in a Bounded Bield of Unknowing.� Cape Town: African Religious Health Assets Programme. Piot P. 2008. “Don't Give up the Fight!� In XVIIth International AIDS Conference, Mexico City, Mexico: UNAIDS. Pisani E. 1999. “Acting Early to Prevent AIDS: The Case of Senegal.� In Best Practice Collection. Geneva: Joint United Nations Programme for HIV/AIDS. Plummer K. 1988. “Organizing AIDS.� In P. Aggleton and H. Homans, eds, Social Aspects of AIDS. London: The Falmer Press. President's Emergency Plan for AIDS Relief (PEPFAR). 2009. “Working with Community-Based and Faith-Based Organisations.� Washington, USA. Rodriguez-Garcia, R., R. Bonnel, N. N’Jie, J. Olivier, F. Pascual, and Q. Wodon. 2011. Analyzing Community Responses to HIV and AIDS: Operational Framework and Typology. Policy Research Working Paper 5532, World Bank, Washington, DC. Rosenberg, C. E. 1992. Explaining Epidemics: And Other Studies in the History of Medicine. Cambridge: Cambridge University Press. Schmid, B., E. Thomas, J. Olivier, and J. R Cochrane. 2008. “The Contribution of Religious Entities to Health in sub-Saharan Africa.� Cape Town: African Religious Health Assets Programme. Seidel, G. 1993. “The Competing Discourses of HIV/AIDS in sub-Saharan Africa: Discourses of Rights and Empowerment vs Discourses of Control and Exclusion.� Social Science & Medicine 36: 175-194. 50 Sider, R., and H. R. Unruh. 2004. “Typology of Religious Characteristics of Social Service and Educational Organizations and Programs.� Nonprofit and Voluntary Sector Quarterly 33: 109-134. Taylor, N. 2005a. “Many Clouds, Little Rain?� The Global Fund and Local Faith-Based Responses to HIV and AIDS. Teddington: Tearfund. _____. 2005b. “The Warriors and the Faithful: The World Bank MAP and Local Faith- Based Initiatives in the Fight Against HIV and AIDS.� Teddington: Tearfund. _____. 2006. “Working Together? Challenges and Opportunities for International Development Agencies and the Church in the Response to AIDS in Africa.� Teddington: Tearfund. _____. 2007. “DFID, Faith and AIDS: A Review for the Update of Taking Action.� London: UK Consortium on AIDS and International Development. TBFF. 2009. “Faith and Malaria: Towards an Integrated Solution.� In Faith and Malaria, Yale University: Tony Blair Faith Foundation, Yale University and Center for Interfaith Action on Global Poverty. Thomas, E., B. Schmid, M. Gwele, R. Ngubo, and J. R. Cochrane. 2006. “Let us Embrace: The Role and Significance of an Integrated Faith-Based Initiative for HIV and AIDS.� Cape Town: African Religious Health Assets Programme. Treichler, P. 1992. “AIDS and HIV Infection in the Third World: A First World Chronicle.� In E. Fee, and D.M. Fox, eds, AIDS: The Making of a Chronic Disease. Berkeley: University of California Press. United Nations Programme on HIV/AIDS (UNAIDS). 2009. Partnership with Faith- Based Organizations: UNAIDS Strategic Framework. Geneva. Weeks, J. 1989. “AIDS: The Intellectual Agenda.� In P. Aggleton, G. Hart and P. Davies, eds, AIDS: SocialRepresentations, Social Practices, New York: The Falmer Press. World Health Organization (WHO). 2004. “The World Health Report 2004: Changing History, Community Participation in Public Health.� Geneva. WHO-CIFA. 2009. “NGO Mapping Standards Describing Religious Health Assets.� Chateau de Bossey Conference Center - Geneva, Switzerland, 10-12 November 2009. Woldehanna, S., K. Ringheim, C.Murphy, et al. 2005. “Faith in Action: Examining the Role of Faith-Based Organizations in Addressing HIV/AIDS - A Multi Country Key Informant Survey. Washington, DC: Global Health Council. 51 CHAPTER 3 MAPPING RELIGIOUS COMMUNITY HEALTH ASSETS AND INITIATIVES: LESSONS FROM ZAMBIA AND LESOTHO Jill Olivier, James R Cochrane and Steven de Gruchy African Religious Health Assets Programme In 2006 a mapping study took place in Zambia and Lesotho in an effort to more systematically understand and describe what role religion had in the health- seeking strategies of local communities, and to map both the formal and informal HIV and AIDS initiatives in the process. That the World Health Organisation (WHO) would commission such research was unusual given long-standing and deep political sensitivities about religion. Also unusual was the deliberate extension of the study beyond standard quantitative geographic information system (GIS) mapping, towards a participatory community engagement model that qualitatively sought out local perspectives on the value and function of the mapped entities for the communities themselves. The results of this study have been reported elsewhere, and its participatory mapping model, described here, is now being utilized in several other countries and contexts. In this note, we address key lessons learned about participatory mapping of religious community responses to HIV and AIDS, focusing on the challenges and strategies of mapping ‘informal’ community-level responses and initiatives. INTRODUCTION Early in the 21st century information gaps on the role and significance of religion to the lives and health-seeking strategies of communities in developing countries were becoming painfully apparent. Equally lacking was information on existing health and development infrastructure supported by faith communities. While the relationship between religion and health at individual level was increasingly well researched, little had been done on community, institutional or population-scale interactions between religion and health. Opportunities for intervention through religious entities seemed potentially significant, but no-one knew what organizational and cultural assets existed that could be built upon. While some estimated, largely anecdotally, that between 20 percent and 80 percent of the medical infrastructure in sub-Saharan Africa was is in the hands of faith-based organizations, at the time of the study (2005-2006) no comprehensive, independent assessment of religious health assets existed, especially in the sub-Saharan region. The few and very limited formal surveys and qualitative assessments that were available demonstrated substantial tangible capacities spanning a broad range of faith traditions and denominations. The World Health Organisation (WHO) therefore, unusually, commissioned a selective mapping study to document the presence and contribution made by religion and religious entities to the struggle for health and wellbeing in Africa, with particular reference to 52 HIV and AIDS. The African Religious Health Assets Programme (ARHAP), a consortium of research partners, 12 undertook the study to identify, map and assess religious health assets (RHAs). Two southern African countries, Zambia and Lesotho, were identified to provide a baseline, both reflecting high levels of poverty, stressed public health systems, and a severe HIV and AIDS impact. The study sought to answer the question: What is the contribution of religion and religious entities to health and wellbeing in the context of HIV and AIDS in Zambia and Lesotho? 13 The task had two related parts. The first was to map health service infrastructure (‘tangible’ religious health assets or RHAs, 14 such as hospitals, clinics, and non-governmental organizations providing health services) held by religious groups, in four sentinel sites in Zambia, and three in Lesotho. Most basically, this aimed to test two hypotheses: that religious communities provide significant health service infrastructure in many parts of Africa; but, that much of it frequently remains ‘hidden’ from national and international eyes and unaligned with national health systems. The second part to the study sought to test a more complex double hypothesis regarding the contribution of religious communities: that they also hold ‘intangible’ religious health assets that can be leveraged for positive health and development outcomes (such as volunteer networks, local wisdom, a sense of coherence and connection, relational systems of community support, trust, and so on); and, that this endows them with value for the communities in which they work. METHODOLOGY, TOOLS, AND PROCESS Five theoretical approaches shaped the mapping study: grounded theory, appreciative inquiry, an integrated mix of quantitative and qualitative tools, participatory disclosure, and interdisciplinary analysis. The two key sets of tools developed to match the theory included GIS mapping (Database and Definitions), and health-seeker and health-provider workshops aimed at a Participatory Inquiry into Religious Health Assets, Networks, and Agency (PIRHANA). GIS mapping Because much of the contribution of religion to public health in Africa is often hidden from view, a quantitative methodology was necessary to identify the real presence of its tangible assets in communities. New computer tools and technologies have accelerated the use of GIS mapping for public health and other purposes. WHO’s Public Health Mapping and GIS Programme has thus produced computerized HealthMapper and 12 WHO’s Partnerships, External Relations, and Communications team, Department of HIV/AIDS entered into a Project Agreement with the Interfaith Health Program, Rollins School of Public Health, Emory University, and collaborating partners of ARHAP - the Universities of Cape Town, KwaZulu-Natal, and Witwatersrand. 13 While Lesotho and Zambia are countries in which Christianity is dominant, this study worked to remain attentive to other religious traditions, including African Traditional Religions (e.g. Traditional Health Practitioners were mapped as ‘religious entities’) 14 ‘religious health assets’ (RHAs) language is part of a broader conceptual framework based on an assets-based logic that partially circumvents the terminology difficulties created by terms such as ‘faith-based organisation’. Tangible RHAs include facilities, care groups, and the like. Intangible RHAs, largely under-recognized, yet often significant, include resilience, hope, coherence, inter-generational support, and other elements of social action and life. 53 Service Availability Mapping (SAM) tools that are now in use by a wide range of partner organizations (see O'Neill and Meert, 2007). For this study, ARHAP’s GIS specialists took the WHO Healthmapper/SAM mapping tools and extended it further, developing a broader coding scheme and category system that, while consistent with the Healthmapper/SAM system, built in a broader range of categories and entities uncovered at a community level relevant to religious entities. Initially ARHAP reviewed and integrated available datasets (from WHO offices, Ministries of Health and Education, land and surveys offices, and others) but, as expected, such data was scarce for the chosen research sites. Nevertheless, where possible, existing provincial or area maps were generated for review. Community Field Workers were employed within the seven sentinel sites in Zambia and Lesotho to collect GPS coordinates and profile sheet information. To capture their diversity, religious entities (REs) were defined according to a four-dimensional matrix: 15 1. Type (noun descriptor): Codes consistent with those already in HealthMapper and SAM databases were used, and matching new codes introduced to identify REs according to their religious-affiliation and primary identity (e.g. Health Provider, Development Agency, Education Provider, etc.; 9 types and 32 subtype codes). 2. Scale and scope, or authority and reach (adjectival descriptor): Including the scale of operations and the location of the seat of authority to be approached in order to leverage the assets held. Primary scope was divided into: Local, District-Regional, National, International. 3. Activities (verb descriptor): Activities were defined by contribution to the challenge of HIV and AIDS (Prevention, Treatment, Care and Support, and Linking/Networking activities). Any one or a combination of 22 identifiers may be attached to a particular religious entity. 4. Time: Locating the entity in time: how long it has been operational, its current stage of organizational development, and major changes over time. 16 “Mapping� as participatory community engagement: PIRHANA The ARHAP study pursued an intentionally integrative approach that blended the ‘facts’ of quantitative mapping with the qualitative and participatory ‘perceptions’ of the local community members and religious-health leaders. For this purpose, an assets-based approach (see Kretzmann and McKnight 1993) was used that, while recognizing the needs and deficiencies that communities face, begins by identifying and building on contextually rooted, locally existing, means to address them. This implies giving priority to what is already present upon which one may build, and that in turn means paying careful attention to the knowledge, experience, networks and resources that local people draw upon in response to their health challenges. This ‘bottom-up’ approach gives 15 Detail of the categorizations and schema is available in the research reports – including category tables. 16 This dimension is not reported on in the original GIS database provided as per the contract requirements. 54 preeminence to community knowledge and perception. The positive contribution of religion and religious entities is foregrounded, though not in denial of the various ways religion might also be harmful to the health of individuals and communities. To meet this integrative approach, a participatory assessment toolset was designed and tested, namely the Participatory Inquiry into Religious Health Assets, Networks, and Agency (PIRHANA). 17 It contributes to the emerging field of Participatory Geographic Information Systems (PGIS), but goes further by holding to the concept of research as community engagement, with a primary commitment to non-extractive research, and the aim of seeking more effective engagement between community, public health services and religious entities. ‘Mapping’ here must therefore be understood in its broadest sense: as a process, using participatory tools, to draw out and diagram community perceptions of religious and health entities and their relationships. PIRHANA was designed to elicit and assess the following: increased understanding of religion and religious entities as health assets, in the local area/context; perceived strengths of the assets; ties and connections between identified assets; changes of scale, order or character among the assets; capacity challenges in relation to the community at large; actual use of the assets by people on the ground; and deeper dynamics in the choices people make about religious health assets, as a crucial contribution to understanding what maps mean in practice, especially in local communities where policies must be implemented (validating the maps on the ground). The PIRHANA toolset consists of two suites of exercises aimed at two levels: community members (health seekers), and regional/national leadership (health- providers). This allows for information and perceptions to be checked and counter- balanced between these different levels. Significant attention is paid to prior participant identification and the invitation process (an intensive on-the-ground purposive sampling process), and onsite preparation. For each level of workshop, participants are selected to represent a predetermined range of demographic and other characteristics. Workshops are carefully facilitated in such a way as to be cognizant of issues of language, power dynamics, and religious, cultural and gender differences. Significant attention is also paid to the logical flow of the exercises, which are designed so that there is coherence to the process itself, as well as to the research findings. For example, community-level exercises flow from contextual considerations to questions about the health and wellbeing within the community context, from the contribution of community facilities to community health and wellbeing to the contribution of religion to health and wellbeing, from the relative contribution of religious entities to health and wellbeing to the identification of exemplars. They end with a process of ‘local commitment’ focused on local strategies. Information from each step is drawn into the exercises that follow. 17 This tool is founded on other bodies of thought and method such as Participatory Rural Appraisal (PRA), and Participatory Learning and Action (PLA); Asset Based Community Development approach (Kretzmann and McKnight, 1993) with the ideas of Appreciative Inquiry (see Elliot 1999). See De Gruchy et al 2011. 55 In 2005 and 2006 a total of 16 Participatory Engagement Workshops were conducted at various locations in the Lesotho and Zambia sentinel sites (which were chosen for rural/urban/regional diversification): 9 community-level and 7 regional-level workshops, including 3 Pilot Workshops in the Copperbelt, Zambia. In total, 358 workshop participants took part. There was a high degree of participation in the workshops and the exercises, and several further community initiatives and collaborative relationships began as a result (although this was not a primary intention of the research, per se). The information generated from these workshops is a combination of quantitative and qualitative data (in the form of maps, spidergrams, timelines, indices, ranked factors, and recorded discussion), generated through a process of anonymous and open participatory disclosure, but always shared publicly. The GIS Mapping and Community Engagement parts were further integrated in the field and later through analysis. For example, local maps were prepared for use in the workshops; and conversely, GIS Fieldworkers were provided with the workshop information for further GIS Mapping work and verification. SUMMARY OF BROADER FINDINGS AND LESSONS LEARNED A large number of previously invisible religious and partner entities were identified and mapped in these community locations as being actively engaged in health and development work, and in particular HIV and AIDS (in the 7 sentinel sites, 494 entities were mapped and collected into a GIS database). Included were large religious hospitals and health facilities; a range of other civil society organizations (FBOs, NGOs, CBOs and the like); as well as a plethora of less formal community-level initiatives and activities. The research highlighted just how limited the available data was, few of the religious entities found having been mapped before. It also confirmed that religious entities and activities are indeed numerous, widespread, highly active, and often invisible to broader national and international view – particularly those functioning at a community level. Community participants themselves consistently emphasized and articulated the more complex ways that such mapped entities should be understood. Researchers were reminded again how religion and religious expression in the African context is often hidden from ‘Western’ view. As one of the researchers noted, it is “like housework in the economy … which is absolutely foundational to economic life yet almost never shows up in standard economic analyses or scholarly debate … religion is so overwhelmingly significant in the African search for wellbeing … so deeply woven in the rhythms of everyday life, and so deeply entwined in African values, attitudes, perspectives and decision-making frameworks that the inability to understand religion leads to an inability to understand people’s lives� (De Gruchy 2006). 56 Summary of key generalized findings 1. Religion is ubiquitous in Zambia and Lesotho, yet often hidden from Western view. Given this, an engagement with religiously informed ‘healthworlds’ is vital for the shaping of public health policy in southern Africa. 2. Religion, health and wellbeing are locally and contextually driven. For those seeking to engage RHAs, religion cannot be viewed as a single, simple cultural variable - no ‘one size fits all.’ 3. Religious involvement in health and HIV/AIDS is increasing - particularly since 2000 - and religious entities have expressed a strong local commitment and desire to be more effective in the area of HIV/AIDS. Interfaith engagement and dialogue require further exploration. 4. Religious entities are perceived as contributing to health, wellbeing and the struggle against HIV/AIDS through tangible and intangible means. It is this combination that distinguishes them and gives them strength. Leading tangible factors comprise compassionate care, material support and health provision; leading intangibles are spiritual encouragement, knowledge giving and moral formation. 5. Certain religious entities are acknowledged as ‘Exemplars’ in the community and these demonstrate exceptional programmatic, operational and associative characteristics. 6. An Assets-Based Approach to research and implementation of religion and health initiatives and HIV/AIDS scale up offers the potential for more rapid, sustainable and effective capacity-building and action. Clearly, the study shows that religion impacts on the health-seeking behaviors of community members in Zambia and Lesotho, often in a complex interweaving of ‘religious’, ‘traditional’ and ‘medical’ strategies. For example, researchers quickly encountered a basic linguistic problem in Lesotho, because ‘religion’ and ‘health’ do not exist as separate terms in Sesotho, while the term that captures both (bophelo) is not immediately translatable within the disaggregated Cartesian discourse that govern standard research questions. In Lesotho, then, addressing questions of health as if they may be separated from larger religious worldviews is flawed conceptually, and will likely lead to flawed interventions. Health in the Basotho context reflects a comprehensive vision of well�being that goes beyond any individual person, to incorporate family, community, nation, ancestors and the earth itself, and religion permeates it all (Germond and Molapo 2006). Consequently, much that is propagated by public health policies and biomedical interventions is frequently dissonant with Basotho views, resulting in miscommunication, with a greater likelihood of failed interventions. Regional variations of the framing role of bophelo as a way of understanding health, religion, and the world were confirmed in other parts of southern Africa. This research does not argue that development and health practitioners must themselves be religious or ‘believers’ in one kind of religion or another, but rather that development and health policy and practice must take seriously such on-the-ground key factors that have a significant impact upon people’s perceptions of health and wellbeing, with a greater willingness to seriously engage with this religiously informed ‘healthworld’ (Germond and Cochrane 2010). 57 MAPPING (INFORMAL) COMMUNITY HEALTH ASSETS The country specific findings having been reported elsewhere, we draw out here some of the findings about the mapping of informal community initiatives. Since the original study, the WHO mapping group has redesigned their HealthMapper and SAM tools, including an additional module inclusive of religious leadership and entities (see WHO- CIFA 2010). But this mapping tool (and others of its sort) continues to place the focus on formal health services, that is, entities with a recognizable structure and organizational capacity, such as hospitals, clinics, NGOs or networks. What remains a great challenge to any systematic mapping and assessment approach is the inclusion of the messy ‘lower’ layer of community initiatives and responses. Because of the inclusion of the participatory inquiry element and the mixed methods approach of the ARHAP study, it was possible for participants to reveal and describe the broader range of informal initiatives occurring in their communities, the bulk of which were not picked up by previous surveys or the local Community Field Workers doing the GIS mapping. These activities and responses to health and development often emerge spontaneously, are not linked to registered ‘non-governmental organizations’, and frequently do use a building or facility that allows for easy GPS mapping. They are rather fluid networks of concerned community actors. For example in Lesotho, local networks of Community Support Groups (CSGs) were consistently identified by communities as being critically important to their well-being. In Lesotho, where access to healthcare services and facilities is beyond the financial reach of ordinary Basotho, there has been a dramatic upsurge of local CSGs. Two different kinds of CSGs were identified: support groups that are served by Village Health Workers linked to formal healthcare facilities, and support groups initiated by local communities that function independently of public healthcare facilities and religious networks. The latter kind were commonly found to be self-initiated, deeply religious (yet not formally linked to any religious structure), made up mainly of female members, and self-funded, using their own resources to feed, cloth, hospitalize, and even medicate patients. These support groups are widely present, are proliferating, and were highly rated by community health seekers as being proactive and addressing the health of the community in practical and effective ways. “We’ve seen a lot of teenagers die…every week, four to five die. When we first met we were three. Then we went to the chief and asked in 2001. By the end of 2002 we said, let’s make a group and go from house to house and ask them about these children... Then the group was ten. We went from house to house getting patients…� (participant comment, Lesotho, 2005) There were many other such examples of exemplar ‘informal’ community initiatives valued by the community such as: a local Islamic community group in Zambia’s eastern region financially supporting a wing of a local government hospital; local health initiatives being run out of religious congregations (often without the knowledge of the religious leader); a religious leader organizing local food parcels for community members who did not have other kinds of social safety nets; a local community group bicycling 58 patients to the local hospital each week; or the significant number of education facilities (schools) engaged in informal health activities. The study demonstrated that formal religious health facilities were present in these communities, and that there had been a dramatic increase of the number of religious NGOs in action since around 2000 (articulated as a response to increased HIV and AIDS funding as well as perceived need). Yet it also showed the wide presence of community level informal interventions and activities, articulated as a response to local need and building on what assets local communities hold and are able to share. It is not possible to know whether such initiatives have increased or decreased in number or impact, as there is no base-line comparison to be made. However, it is significant to recognize that these community responses (many of them sustained efforts) continue, irrespective of whether they are recognized or linked to national and international plans and systems. They are rarely acknowledged in databases, nor is there sufficient case-study material available to understand how such responses are initiated, evolve and change over time, or might be sustainably supported without destroying their essential nature. The study thus recommended: the further use of mapping in strategic ways to recognize the assets on the ground and their potential connections; the need to strengthen local community support groups working in the field of health and wellbeing, and link them to public health structures; the need to support the work and replication of ‘network hubs’ (or intermediaries) to leverage existing religious health assets and community initiatives, and to nurture additional ones; and the need to further link to the exemplar entities as identified by the communities in this study to understand promising practices and implications. PARTICIPATORY MAPPING AS COMMUNITY ENGAGEMENT Participatory mapping carries both inherent strengths and limitations. While it allows for the addition of a layer of qualitative perception, it also makes ‘mapping’ more challenging, especially for those accustomed to even GIS data at a national scale. As is evident, the entire research process was driven by a participatory, inductive and non- extractive approach. The use of group-identified participatory factors in all the ranking exercises and the contextual identification of social and religious facilities and entities generated data that is highly context-specific, and the generation of research data is therefore uneven across the research sites. While appreciating local wisdom and practice is greatly enriching, participatory mapping is less stream-lined, more costly, and more time-consuming than many other methods. Said differently, participatory mapping can be a useful community engagement tool, but translating bottom-up community perspectives into top-down policy remains a challenge. This includes finding adequate tools and research methodologies to respond to the information needs of policy- and decision- makers, while simultaneously responding to the need to better understand the complexity of how religious health assets and religious entities act and are understood in the communities in which they are based. 59 Nevertheless, ARHAP’s experience is that appreciative ‘mapping’ helps to empower local communities to see their own assets and potentialities, to create ‘ideal’ maps of the future, and to create a powerful communication tool for planning and decision-making (if used appropriately). This has also been shown in various modifications of PIRHANA- style participatory asset mapping and assessment tools for a variety of needs and settings in Africa, Europe and in the USA, where participatory mapping methods have enabled an engagement with the complex nature of communities on the ground, including their religiously motivated and linked initiatives and perspective, with measurable positive outcomes. Community maps are fluid – they are moving and living maps – with indistinct data points in changeable landscapes. Informal community initiatives emerging through communities of faith are particularly complex, with few generalizable standards. It is not possible to force an overly simplistic view on the deeply textured study results. Most importantly, the approach confirms that religion, health, and well-being are contextually driven, and that local context is most significant of all. Ordinary Zambians and Basotho perceive their daily struggle for health and well-being and their response to HIV and AIDS in the overall context of a daily struggle for survival marked by extreme poverty, weak public health capacity, and serious environmental challenges, as well as factors in their local communities. Similarly, their religious belief systems and practices are locally, contextually driven. There is a rich variety of religious expression, as evidenced by the diversity of participants and religious entities included here and by the underlying influence of African traditional beliefs and healing practices. There is no ‘one size fits all’ approach. The findings of the ARHAP study reinforce previous recommendations that policies, interventions, and resource allocation be driven by local strategies and that local assets and agency be appreciated and strengthened. Furthermore, in the dialogue between religion and health this means that just as there is no such thing as ‘health’ other than health for particular people in a particular context, so there is no such thing as ‘religion’ other than a particular religious expression by particular people in a particular context. Religious landscape and motivations are complex and must be appreciated and understood as such. Alignment with public policy requires one to deal with these assets in many different ways: as one of our field researchers put it, “You can’t talk to one bishop and think you’ve dealt with the whole religious community.� These contextual realities have significant implications for health interventions and for replicability of research and programmatic approaches and strategies. 60 REFERENCES Abbot, J., R. Chambers, C. Dunn, T. Harris, E. Merode, G. Porter, J. Townsend, D. Weiner, and E. de Merode. 1998. “Participatory GIS: Opportunity or Oxymoron?.� PLA Notes 33, IIED: London. African Health Assets Programme (ARHAP). 2006. Appreciating Assets: The Ccontribution of Religion to Universal Access in Africa. Cape Town: African Religious Health Assets Programme. Chambers, R. 2006. “Participatory Mapping and Geographic Information Systems: Whose Map? Who is Empowered and Who is Disempowered? Who Gains and Wwho loses?.� EJISDC 25(2): 1-11. De Gruchy, S. 2006. “Like Housework in the Economy, the Hidden Ubiquity of Religion in African Well-Being.� Presentation at Atlanta: Religion and Health Connection, Rollins School of Public Health, Emory University. De Gruchy, S., J. R. Cochrane, J. Olivier, and S. Matimelo. 2011. "Participatory Inquiry on the Interface Between Religion and Health: What Does it Achieve, and What Not?." In J. R. Cochrane, B. Schmid, and T. Cutts, eds, When Religion and Health Align: Mobilising Religious Health Assets for Transformation, Cluster Publications: Pietermaritzburg. Elliot, C. 1999. Locating the Energy for Change: An Introduction to Appreciative Inquiry. Winnipeg: International Institute for Sustainable Development. Germond, P., and J. R. Cochrane. 2010. “Healthworlds: Conceptualizing Landscapes of Health and Healing.� Sociology 44 (2):307-324. Germond, P., and S. Molapo. 2006. “In Search of Bophelo in a Time of AIDS: Seeking a Coherence of Economies of Health and Economies of Salvation.� Journal of Theology for Southern Africa 126:27-47. Kretzmann, J. P., and J. L. McKnight. 1993. Building Communities From the Inside Out: A Path Toward Finding and Mobilizing a Community's Assets. Chicago: ACTA Publications. O'Neill, K., and J. P. Meert. 2007. Putting People and Health Needs on the Map. Geneva: World Health Organization. WHO�CIFA. 2010. “NGO Mapping Standards Describing Religious Health Assets.� Chateau de Bossey Conference Center � Geneva, Switzerland, 10�12 November 2009. 61 CHAPTER 4 MAPPING RELIGIOUS HEALTH ASSETS: ARE FAITH-INSPIRED FACILITIES LOCATED IN POOR AREAS IN GHANA? Harold Coulombe and Quentin Wodon The World Bank The Christian Health Association of Ghana is the largest federation of faith- inspired provider of health services in Ghana. With 168 hospitals and clinics at the time this paper was written, 70 percent of which are affiliated with the Catholic Church, the federation accounts for more than a fourth of all hospital beds in the country according to administrative data from the Ministry of Health. Using poverty mapping techniques and simple regression analysis, this paper aims to answer the following question: Are CHAG facilities located primarily in areas that have a high proportion of Catholics or Christians, or in areas that have a high proportion of the population living in poverty? It appears that the location of CHAG facilities is correlated today more with the share of Catholics living in specific districts than with the level of poverty in those districts. INTRODUCTION There is substantial interest in developing countries in mapping religious health assets, with mapping exercises taking many different forms. Some analysts rely on GIS technology to locate formal health facilities, with the aim of guiding the location of new facilities, as well as understanding potential overlaps or complementarities among facilities from different providers but located within a given area. The World Health Organization has undertaken work in this area among others through its Service Availability Mapping (SAM) project (see also WHO-CIFA 2009). In Ghana SAM provided detailed information on the distribution of health services at the district level, including in the area of HIV-AIDS baseline monitoring data (Ministry of Health 2007). Some private providers of health services also conduct mapping analysis of this type, albeit in a simpler way, to show their contribution to national health systems. For example the Christian Health Association of Ghana (CHAG 2009) collected information on all the facilities operated by its institutional members affiliated with various Christian denominations. Geographic mapping should be differentiated from another type of mapping exercises which is more participatory in nature and focuses on documenting and understanding the role of informal groups and organizations in specific aspects of health care provision or support (see for example Olivier et al 2012, based on ARHAP 2006, see also de Gruchy et al 2011). These approaches can be useful to locate geographically specific community- based activities, especially when they integrate GIS data, but the primary objective is often more to gather data on community perceptions and servicers as part of a broader process of understanding and community engagement in health. Still other ‘mapping’ 62 studies aim simply to document the role of various actors in the health system of a country. These exercises tend to take the form of reviews of existing literature or information, at times accompanied by additional qualitative interviews, and could be considered as exercises in ‘landscaping’ the often broad range of organizations involved in a country’s health system. Such studies help in showing the role of non-governmental organizations in health beyond the traditional focus on the operation of public facilities operated by Ministries of Health. In this paper, within the broader literature on mapping religious health assets, our objective is to assess whether faith-inspired health service providers in Ghana are located in the country’s poor areas. The focus is on the Christian Health Association of Ghana, the largest faith-inspired provider of health services in Ghana. The number of CHAG Member Institutions or facilities has grown from 25 in 1967 to 168 in 2008, with approximately 70 percent of the institutional members linked to the Catholic Church. In terms of regional distribution, the Ashanti region has the highest number of CHAG facilities (23 percent of the total), followed by the Brong Ahafo and Eastern regions (12.5 percent each) and the Northern region (11 percent). Other regions have only between three and ten percent of CHAG facilities (on Christian Health Associations in Africa more generally, see Dimmock et al 2012; see also Wodon 2013). In order to assess whether CHAG facilities are located in poor areas, we combine administrative data on the location of the facilities with data from a new poverty map constructed for Ghana by Coulombe and Wodon (2012a). Following work by Elbers et al (2002, 2003) who have shown how to construct detailed poverty maps by combining census and household survey data, there has been a growing literature on the construction of these maps and their use. In Ghana a first poverty map was constructed by Coulombe (2008) using the fourth round of the Ghana Living Standards Survey (GLSS4) implemented in 1998/99 and the Housing and Population Census of 2000. Yet that map probably fails to represent the geography of poverty today because poverty has been reduced dramatically from 39.5 percent in 1998-1999 to 28.5 percent in 2005-2006 according to results based on the fifth round of the Ghana Living Standards Survey (Coulombe and Wodon 2007). Furthermore, the reduction in poverty has not been uniform in the country. The data suggested a possible increase in poverty in the capital city of Accra, a sharp reduction in poverty in the coastal and forest areas, and stagnation or only very limited progress towards poverty reduction in the northern savannah area. These changes in the geography of poverty must be taken into account in order to assess the extent to which today CHAG facilities are located in poor areas. The structure of the paper is as follows. Section 2 provides information on our data sources, which include household survey as well as CHAG and census data. In section 3, we try to answer two questions. First, we ask whether in a simple econometric model CHAG facilities tend to be located in areas with a higher proportion of Catholics or in poorer areas controlling for the correlation between the two types of areas. Second, independently of the faith affiliations in various districts, we ask statistically whether facilities are located proportionately more in poor areas or not. A conclusion follows. 63 DATA AND METHODOLOGY Four sources of data are used in the paper. The first two sources are nationally representative household surveys. The first survey is the Ghana Living Standards Survey (GLSS5) implemented in 2005-2006. The GLSS is a multi-purpose household survey covering demography, health, education, employment, migration, housing, agriculture activities, non-farm self-employment, household expenditures, durable goods and, remittances and other incomes. The 2005/06 round of the survey was administrated to around 36,500 individuals grouped into 8700 households. This nationwide sample is deemed representative at the level of the ten regions. The second survey is the large sample (50,000 households) 2003 Core Welfare Indicator Questionnaire (CWIQ) survey, deemed representative at the district level. Although this second survey does not provide information on the consumption level and thereby poverty of households, it has the advantage of being representative at the district level due to its large sample size. Data from both household surveys are used to construct in the large sample CWIQ survey a poverty map that is representative at the district level. The poverty map was estimated by Coulombe and Wodon (2012a) using the methodology developed by Elbers et al (2002, 2003). The basic idea is simple, but its implementation is complex. The large sample CWIQ survey is used as an equivalent to a census – while it does not include consumption data, it is large enough in terms of sample size so that it can be used to estimate district-level poverty measures. In practice, consumption data for each household is imputed into the CWIQ using the poverty mapping technique. To this end, a regression of adult equivalent consumption is first estimated using the GLSS5, limiting the set of explanatory variables to those common to both the GLSS5 survey and the CWIQ (and applying a test of the equality of the means of all selected explanatory variables to ensure comparability between the two surveys). Next, the coefficients from that regression as well as the structure of its error terms are applied to the CWIQ data to predict consumption for each household in the CWIQ. Third, predicted consumption in the CWIQ is used to estimate district level poverty measures which tend to be precise at that level of aggregation because of the sufficient number of observations on which the measures are based 18. The details of the construction of the CWIQ-based poverty map are given in Coulombe and Wodon (2012a). The result is a series of poverty estimates at the district level which can be used to assess whether faith-inspired facilities are located in poorer or wealthier districts. At the time of the implementation of the CWIQ survey in 2003, there were 110 18 Although the above procedure is conceptually simple, its implementation requires complex computations due to the need to take into account spatial autocorrelation (expenditure from households within the same cluster or area are often correlated) and heteroskedasticity in the development of the predictive model. Another issue is the need to compute standard errors to assess the degree of precision of poverty estimates. Those standard errors are important since they help assess how far the information can be disaggregated (the smaller the area and the number of observation are, the larger the standard errors of the poverty measures are likely to be), and whether using a large survey can be a legitimate alternative to using a census. 64 districts. In 2004, a district remapping yielded 28 new districts, essentially by splitting a number of large districts into two separate districts (or in one case by combining two adjacent districts and splitting them into three districts). The estimations of poverty used here are those based on the original 110 districts, as this is the level at which the CWIQ survey is deemed representative. The poverty map obtained for Ghana is visualized in the first map on the left of Figure 1 for the headcount index of poverty. Clearly much higher measures of poverty are observed in the northern parts of the country. In addition to these two surveys used to construct the poverty map for Ghana, we also use administrative data on the location of all CHAG. The number of CHAG Member Institutions of facilities has grown from 25 in 1967 to 168 in 2008. Data on CHAG facilities by district available according to the new definition of those districts (post- 2008) were mapped to the 110 districts available in the CWIQ. The 168 CHAG facilities, including 75 clinics, 3 specialist clinics, eight health centers, 58 hospitals, 13 primary health care units, one polyclinic, and ten training institutes. Table 1 below shows that the Ashanti, Brong Ahafo and Eastern regions had the largest number of facilities in 2008. Of the 168 facilities, information on the number of beds was also available for 120 facilities, and of those 77 facilities had beds (table 2). The data can be used to compute the number of CHAG beds per 1,000 individuals at the district level, which is probably a better way to assess the allocation of resources than relying solely on the number of facilities (given that facilities with a larger number of beds tend to provide substantially more services than other facilities). Table 1: Regional Distribution of CHAG Facilities Share Region Hospital Clinic PHC TI HC Polyclinic Total of total (%) Ashanti 17 18 1 3 39 23 Brong Ahafo 10 6 2 2 1 21 12.5 Central 4 6 10 6 Eastern 6 11 2 1 1 21 12.5 Greater Accra 2 3 1 6 3 Northern 5 9 2 2 18 11 Upper East 1 8 3 1 3 16 10 Upper West 2 1 1 2 6 3 Volta 7 6 1 1 15 9 Western 4 10 1 1 16 10 Total 58 78 13 10 8 1 168 100 Source: CHAG (2009). Note: PHC = primary health care units; TI = training institutes; HC = health centers. Table 2: Information on Number of Hospital/Clinic Beds for CHAG Facilities Type of facilities Number of facilities Number of facilities Number of facilities with information on with positive number of beds number of beds Clinic 75 53 26 Clinic specialist 3 2 0 Health center 8 5 4 Hospital 58 50 44 PHC (Primary Health Care) 13 9 2 Polyclinic 1 1 1 Training institute 10 0 0 Total 168 120 77 Source: CHAG (2009). 65 Finally, the fourth and last source of data is Ghana’s last census (dating back to the year 2000) which is used to gather information on the share of individuals affiliated with particular faiths at the district level. Because approximately 70 percent of the facilities operated by CHAG are Catholic, we focus on this denomination in the empirical work in the next section. A visualization of the various sources of data is provided in Figure 1. The first map provides a visualization of poverty by district according to the headcount index which measures the share of the population that is poor in each district. This map is obtained using the first two sources of data mentioned above, namely the GLSS5 survey and the CWIQ survey. The poverty map was obtained by applying the technique of the poverty map using the two surveys, with the model for the correlates of poverty estimated with the GLSS5 data and applied to the data in the CWIQ survey in order to come up with district level estimates of poverty. The second map is based on the administrative data from CHAG, which were used together with population data by district in order to compute bed rates in each of the district. The last map provides the share of the population who are Catholics according to the 2000 census. Clearly, when comparing the three maps, it appears that some of the areas where CHAG has a higher density of facilities as measured through bed rates are also areas that tend to have a higher share of the population that is Catholic, while the relationship with poverty measures appears to be weaker. In the next section simple regression analysis is used to look more closely at these relationships. 66 Figure 1: District-level Poverty Headcount and CHAG Bed Rate in Ghana Headcount Index (%) CHAG Bed Rate, per 1000 Share of Catholics (%) Sources: Authors’ calculation based on the 1998 Census, the GLSS 2005-2006 and the CWIQ 2003 survey data, as well as administrative data from CHAG. 67 RESULTS In this section, we would like to answer two simple questions. The first question is whether CHAG facilities tend to be located primarily in areas that have a higher proportion of Catholics, or in areas that have a higher proportion of the population living in poverty. This is analyzed using simple econometric techniques. The second question is whether CHAG facilities are located proportionately more in poor areas, independently of whether this is related or not to the proportion of Catholics in those areas. This is analyzed through statistical analysis, considering both the districts in which CHAG has facilities and all districts taken together, including those where CHAG is not present. To answer the first question, we estimate six simple regressions. The first regressions are probits to assess the correlation between the presence or absence of any CHAG facility in the various districts as a function of four variables: the share of the population in the district that is urban (as a proxy for the need for facilities such as hospitals and clinics that tend to be located in or near urban centers), the share of the population that is Catholic, the share of the population that is associated with other Catholic denominations, and finally the share of the population in poverty. The probit regressions are estimated both for CHAG facilities as a whole, and for the facilities that belong to the Catholic Church. Next, we estimate ordinary least squares (OLS) regressions in those districts that have a positive number of beds from CHAG facilities to relate the bed rate ratio to the same four variables. Finally, we estimate censored regressions (tobits) that factor in both the zero values and the positive values for the bed rates in a single estimation. In all regressions, the only variable that appears to be statistically significant is the share of the population that is Catholic (see table 3). Even if the magnitude or size of the coefficients seems small, and even if the explanatory power of the regressions is also small, this nevertheless suggests that the location of CHAG and especially Catholic facilities was probably driven (historically) more by the share of the population that was Catholic or Christian in various parts of the countries than by poverty. We must however acknowledge that while the faith landscape of various areas is likely to changes only relatively slowly over time (despite the growth of the Charismatic movement), poverty measures can change more rapidly within one or two decades, and we use here contemporary measures of poverty (representative of the year 2005-2006) rather than those that prevailed when most CHAG facilities were established over thirty years ago. 68 Table 3: Correlates of the CHAG Hospital Bed to Population Ratio in Ghana Probit (dF/dX) OLS for Positive Tobit Bed Rates All Catholic All Catholic All Catholic Constant .. .. -0.0900 -0.0439 -118.1 -114.1 Urban (%) -0.0028 *0.0040 0.0002 0.0008 0.6165 -0.7155 Catholic (%) **0.0113 **0.0104 **0.0029 ***0.0041 ***4.4591 **5.2353 Other Christian (%) 0.0005 -0.0017 0.0012 0.0005 0.5180 0.5724 Poverty (%) -0.0037 -0.0008 0.0181 0.0008 0.0705 -1.1577 R2 or Pseudo-R2 0.080 0.046 0.136 0.159 0.010 0.013 Observations 110 110 57 39 110 110 Source: Authors’ estimation. Note: Levels of statistical significance: *:0.90; **:0.95; ***:0.99. Our results are limited by the fact that we do not have data here on the precise location of the CHAG facilities within districts that can be combined with equally precise (and current) data on poverty at sub-district levels. This is because the poverty map used can be considered as statistically robust only at the level of districts, and the previous poverty map estimated by Coulombe (1998) does not represent the situation of poverty in Ghana today. It has been argued that CHAG facilities tend to be located more in poor and marginalized areas within districts than is the case for public facilities. While we can’t check this directly here, the household survey evidence to this effect is however rather weak. As noted by Coulombe and Wodon (2012b) one observes in national household surveys few differences between faith-inspired and public facilities in terms of how they serve different households groups defined according to quintiles of well-being, and in addition the benefit incidence of both faith-inspired and public facilities still remains to tilted towards better off segments of the population. This does not seem to suggest then that on average, CHAG facilities would be serving within a given district the poor substantially more than other groups, or in comparison with public facilities. Now, even if it is correct that historically the location of CHAG facilities was driven in part by where Catholics and other Christians lived, this does not mean that the facilities do not serve the poor today. A key aspect of this question remains whether CHAG facilities are located proportionately more in poor areas than in wealthier areas. In table 4, the facilities are grouped by quintiles of district-level well-being (ranked from the poorest to the richest districts, with each quintile representing one fifth of all districts), as well as by quintiles of the population assuming that in any district, a CHAG facility would serve proportionately in the same way all individuals from various levels of well-being living in that district (again, this assumption does not appear to be too restrictive given the results on benefit incidence analysis using household surveys obtained by Coulombe and Wodon 2012b in this volume). The results in table 4 suggest that when looking at quintiles of districts, CHAG facilities tend to serve slightly more districts that have a higher level of well-being on average, while by quintiles of population (taking into account the size of the population in each district, which is more appropriate), CHAG facilities tend to serve slightly more the poorer quintiles. The evidence as to whether CHAG facilities are located proportionately more in poor areas is thus somewhat mixed, but overall the pattern seems to reflect a 69 broad distribution of the facilities in the country in both poor and less poor areas, so that this tends to confirm the results of benefit analysis using household surveys (Coulombe and Wodon 2012b). In table 4, the data is provided in terms of facilities rather than beds, but the results are similar when considering the number of beds. Table 4: Number of CHAG Facilities, by Quintiles and Type, 2009 Poorest Second Third Fourth Richest Total Quintile of Individuals Hospitals 10 18 15 9 7 59 Clinics 31 11 35 16 6 99 Training Institutes 3 1 2 3 1 10 Total 44 30 52 28 14 168 Quintile of Districts Hospitals 6 11 17 12 13 59 Clinics 26 8 23 25 17 99 Training Institutes 3 0 1 5 1 10 Total 35 19 41 42 31 168 Source: Authors’ calculations based on qualitative fieldwork data. Notes: The quintiles of individuals represent 20 percent of individuals living in districts ranked by the share of their population in poverty; the quintiles of districts represent 20 percent (i.e., 22 districts in each case) of the districts ranked by the share of their population in poverty. In Figures 2 and 3, data on the number of beds instead are used. The figures provide scatter plots of on the vertical axis the number of beds in CHAG facilities per 1,000 inhabitants in the various districts, while the horizontal axis represents either the share of the district population in poverty (in Figure 2) or the average level of consumption per equivalent adult in the district (in Figure 3). In each Figure, four scatter plots are provided. The first two scatter plots include all districts, including those in which CHAG does not have beds. The difference between the two scatter plots is that in the second one, rather than relying only on the number of beds, we also factor in the regional level estimates of occupancy rates for facilities with bed computed by CHAG for its facilities. As noted by CHAG, occupancy rates in CHAG facilities in 2006 varied from 42 percent in the Upper West (and poor) region to 4.9 percent in the Western (and better off) region. The size of the dots in the scatter plots are proportional to the size of the districts in terms of population, weighted in the second scatter plot by the regional-level occupancy rates for the CHAG facilities. Linear and quadratic trend lines are then fitted through the scatter plots using weighted regressions, with the weights being proportional the size of the dots (that is, proportional to the district population with or without occupancy rates adjustment). In Figure 2, upward trend lines would suggest that CHAG facilities are proportionately serving the poor more than the non-poor, at least in terms of their location. The analysis is also carried only on those districts that have a positive number of beds in CHAG facilities, although for national representativeness it is in principle to include all districts in the estimations of the trend lines. The trend lines suggest that there is some mild indication that the location of CHAG facilities is indeed pro-poor (especially when looking only at the districts where CHAG has facilities), but this ‘pro-poorness’ tends to be relatively weak. The issue of whether one should use the scatter plots with or without 70 the districts where CHAG has facilities may warrant further discussion. From the point of view of CHAG which may look only at the distribution of its facilities in the districts where it is indeed active, the message may be that there is quite a significant bias towards poor areas. But from the point of view of a national government or an outside observer who looks at the distribution of beds in the country as a whole, that relationship is much weaker, in part because CHAG does not have facilities in many relatively poor districts. Figure 2: CHAG Beds per Thousand Persons by District and Poverty Headcount All District All Districts Not Weighted Weighted by Occupancy Rates 0.15 0.3 0.10 0.2 0.05 0.1 0.00 0.0 0.0 20.0 40.0 60.0 80.0 0.00 20.00 40.00 60.00 80.00 Poverty Heacount (in %) Poverty Heacount (in %) Bed rate (in %) Linear fit Combined bed/occupacy rate (in %) Linear fit Quadratic fit Quadratic fit District with CHAG Beds District with CHAG Beds Not Weighted Weighted by Occupancy Rates 0.15 0.3 0.10 0.2 0.05 0.1 0.00 0.0 0.0 20.0 40.0 60.0 80.0 0.00 20.00 40.00 60.00 80.00 Poverty Heacount (in %) Poverty Heacount (in %) Bed rate (in %) Linear fit Combined bed/occupacy rate (in %) Linear fit Quadratic fit Quadratic fit Source: Authors’ estimation based on GLSS5 2005-2006, CWIQ 2003, and CHAG data. The same result is observed in Figure 3, where instead of looking at the relationship between the bed rates and poverty, we use instead consumption per equivalent adult to measure well-being. The trend lines are downward sloping, suggesting that CHAG facilities tend to be located in areas with lower levels of consumption, but again the relationship is weak unless one considers only districts with CHAG facilities. Overall then, there is some pattern in the scatter plots that suggest some relationship between the placement of facilities and poverty or consumption at the district-level, but the 71 relationships are not very strong when the whole sample of districts is taken into account, and the relationships tend to be driven at least in part by high levels of poverty in a few northern districts with a high share of the population affiliated with the Catholic faith. Figure 3: CHAG Beds per Thousand Persons by District and Consumption Level All District All Districts Not Weighted Weighted by Occupancy Rates 0.15 0.3 0.10 0.2 0.05 0.1 0.00 0.0 0 100 200 300 400 0 100 200 300 400 Per Equivalent Adult Consumption (in GHC) Per Equivalent Adult Consumption (in GHC) Bed rate (in %) Linear fit Combined bed/occupacy rate (in %) Linear fit Quadratic fit Quadratic fit District with CHAG Beds District with CHAG Beds Not Weighted Weighted by Occupancy Rates 0.15 0.3 0.10 0.2 0.05 0.1 0.0 0.00 0 100 200 300 400 0 100 200 300 400 Per Equivalent Adult Consumption (in GHC) Per Equivalent Adult Consumption (in GHC) Bed rate (in %) Linear fit Combined bed/occupacy rate (in %) Linear fit Quadratic fit Quadratic fit Source: Authors’ estimation based on GLSS5 2005-2006, CWIQ 2003, and CHAG data. These results on the relative weakness in the relationship between the location of CHAG facilities and areas with high poverty or low levels of consumption when all districts are taken into account may seem surprising given the widespread perception that CHAG serves primarily rural and poor areas. Yet it is also recognized that even if historically CHAG facilities were indeed located several decades ago in underserved areas, patterns of migration and development may have changed the nature of the areas in which CHAG facilities are located. As stated by CHAG (2006) itself in one of its annual reports, “a few [facilities] are in big towns now but were built there when the towns were small and rural. A few can now also be seen in the slums of some of the cities. These are targeted at serving the health needs of the poor and vulnerable populations that have been created 72 by urbanization.� In addition, there have been profound changes in the geography of poverty in the country over the last two decades, with poverty being increasingly concentrated in the northern and rural savannah part of the country (Coulombe and Wodon 2007). Given that there are generally more CHAG clinics and hospitals in the southern and middle belts than in the northern areas, albeit with the exception of a few districts in the Upper West region where CHAG has important facilities, the changing patterns of poverty may have reduced the proportion of the services provided by CHAG that is allocated to some of the poorer areas of the country. In addition, while the share of CHAG beneficiaries located in poor areas may have been reduced, new public facilities such as clinics, health posts and mobile programs providing community-based health services have been expanded by the Ministry of Health in rural areas for primary health care (Salisu and Prinz 2009). Together these three phenomena - population growth, migration, and development in traditional CHAG areas; a higher concentration of poverty in some of the northern regions where CHAG has traditionally had a smaller presence; and an expansion of public services by the Ministry of Health in underserved areas over the last decade or two – may have led to a reduction of the footprint of CHAG hospitals and clinics in some of the poorer areas of the country as compared to the footprint of public facilities. These results on the location of facilities are consistent with findings from the analysis based on household surveys that suggests that faith-inspired providers, most of which belong to CHAG, may not today serve the poor much more than public providers. CONCLUSION Using poverty mapping techniques as well as simple statistical and regression analysis, the objective of this paper was to answer two questions. First, are CHAG facilities located primarily in areas that have a high proportion of Catholics or Christians, or in areas that have a high proportion of the population living in poverty? To the extent that this can be measured with contemporary data, it appears that the choice of the location of CHAG facilities may have been driven more by the share of Catholics or Christians living in specific districts than by the level of poverty in those districts. This would not be a surprising finding given the history of mission activities in Ghana, and it is not a problem in itself, so long as all of those who live in certain areas may benefit from the services provided by facilities, which is certainly the case. Second, independently of whether the answer to this question is related or not to the proportion of Catholics or Christians living in any given area, are CHAG facilities located proportionately more in poorer than in better off areas? The results provide mild evidence that CHAG facilities tend to be located in somewhat poorer areas. From the point of view of CHAG which may look only at the distribution of its facilities in the districts where it is active, the evidence that it tends to serve poorer areas is stronger. But from the point of view of a national government or an outside observer who looks at the distribution of beds (and facilities) in the country as a whole, the relationship is weaker, in part because CHAG does not have facilities in many relatively poor districts. As to why the bias of CHAG towards poor areas seems lower than one might have expected, at 73 least three explanations could be provided. Faster development in traditional CHAG areas, a higher concentration of poverty in some of the northern regions with substantial Muslim populations where CHAG has traditionally had a smaller presence, and an expansion of public services by the Ministry of Health in underserved areas over the last decade or two (although this was not measured here) may all have led to a reduction of the role that CHAG has historically played in preferentially serving the health care needs of the poor in Ghana. REFERENCES African Health Assets Programme (ARHAP). 2006. Appreciating Assets: The Contribution of Religion to Universal Access in Africa. Cape Town: African Religious Health Assets Programme. Blevins, J., S. Thurman, M. Kiser, and L. Beres. 2012. “Community Health Assets Mapping: A Mixed Method Approach in Nairobi.� In J Olivier and Q Wodon, eds, Mapping, Cost, and Reach to the Poor of Faith-inspired Health Care Providers in sub- Saharan Africa, World Bank, Washington, DC. Christian health Association of Ghana (CHAG). 2006. Annual Report: June 2005 - May 2006. Accra. ____. 2009. Christian Health Association of Ghana: 2008 Annual Report. Accra: Christian Health Association of Ghana. Dimmock, F., J. Olivier, and Q. Wodon. 2012. “Half a Century Young: The Christian Health Associations in Africa.� In J. Olivier and Q. Wodon, eds, The Role of Faith- Inspired Health Care Providers in sub-Saharan Africa and Public-private Partnerships. World Bank, Washington, DC. Elbers, C., J. O. Lanjouw, and P. Lanjouw. 2002. Welfare in Villages and Towns: Micro Level Estimation of Poverty and Inequality, Policy Research Working Paper No. 2911, World Bank, Washington, DC. _____. 2003. Micro-Level Estimation of Poverty and Inequality. Econometrica 71(1), 355-364 Coulombe, H. 2008. “Ghana Census-Based Poverty Map: District and Sub-District Level Results.� In E. Aryeetey and R. Kanbur, eds, The Economy of Ghana: Analytical Perspectives on Stability, Growth and Poverty. Oxford: James Currey. Coulombe, H., and Q. Wodon. 2007. “Poverty, Livelihoods and Access to Basic Services in Ghana.� In World Bank, Ghana: Meeting the Challenge of Accelerated and Shared Growth (country economic memorandum). Report No. 40934-GH, Volume III: Background Papers, Washington, DC. 74 _____. 2012a. “Poverty Maps Without Census Data? The Case of Ghana, Mimeo.� World Bank, Washington, DC. _____. 2012b. “Benefit Incidence Analysis of Public Health Spending in Ghana.� In J. Olivier and Q. Wodon, eds, Mapping, Cost, and Reach to the Poor of Faith-inspired Health Care Providers in sub-Saharan Africa. World Bank, Washington, DC. De Gruchy, S., J. R. Cochrane, J. Olivier, and S. Matimelo. 2011. “Participatory Inquiry on the Interface between Religion and Health: What Does it Achieve, and What Not?.� In J. R. Cochrane, B. Schmid, and T. Cutts, eds, When Religion and Health Align: Mobilising Religious Health Assets for Transformation. Cluster Publications: Pietermaritzburg. Ministry of Health. 2007. Service Availability Mapping (SAM). Accra: Republic of Ghana. Olivier, J., J. R. Cochrane, and S. De Gruchy. 2012. “Mapping Religious Community Health Assets and Initiatives: Lessons from Zambia and Lesotho.� In J. Olivier and Q. Wodon, eds, Mapping, Cost, and Reach to the Poor of Faith-inspired Health Care Providers in sub-Saharan Africa. World Bank, Washington, DC. Salisu, A., and V. Prinz. 2009. Healthcare in Ghana. Wien: Austrian Centre for Country of Origin and Asylum Research and Documentation, and Austrian Red Cross. WHO�CIFA. 2009. NGO Mapping Standards Describing Religious Health Assets. Chateau de Bossey Conference Center, Geneva, Switzerland, 10�12 November. Wodon, Q. 2013. Faith and Human Development: Education and Health Services in Africa, New York: Palgrave MacMillan. 75 CHAPTER 5 DIFFERENCES IN THE PRIVATE COST OF HEALTH CARE BETWEEN PROVIDERS AND SATISFACTION WITH SERVICES: RESULTS FOR SUB-SAHARAN COUNTRIES Clarence Tsimpo and Quentin Wodon The World Bank The issue of whether faith-inspired providers are able to reach the poor depends in part on the cost of the health services provided. This paper relies on recent nationally representative household surveys for sub-Saharan African countries to assess to what extent the cost of healthcare is a major reason for not being satisfied with health services and whether concerns with costs differ between types of providers. The paper also provides estimates of the cost of healthcare in a half dozen countries, again comparing public, private secular, and faith- inspired providers. The results suggest that cost indeed remains a major concern for households. There are differences in out-of-pocket costs for households between providers, with in many cases public providers being cheaper than faith- inspired providers and private secular providers. Yet these differences depend on the country and are not as large as one might have assumed. INTRODUCTION Whether faith-inspired providers are able to reach the poor depends in part on the cost of the health services that they provided. It is often believed that faith-inspired providers in Africa do reach the poor in part because they make special efforts to make their services affordable to them at low cost. This concern for the poor and the affordability of care is itself related to the fact that Africa’s two leading religions, Christianity and Islam, both have longstanding traditions of service to the poor, including in the area of healthcare. There is some evidence that faith-inspired health facilities have altruistic motives, or at least do not behave in the way that for-profit health providers would, and this has implications for cost recovery of the services provided. In their analysis of health service provision in Uganda, Reinikka and Svensson (2010) used a change in financing of not- for-profit health care providers through untied grants to test two theories of organizational behavior. The first theory postulates that not-for-profit providers are intrinsically motivated to serve the poor and will therefore use new resources to expand their services or cut the cost of these services. The second theory postulates that not-for- profit providers are captured by their managers or workers and behave like for-profit actors. Although they may not appropriate profits, they would tend to use untied grants to raise the salaries of their staff or provide them with other benefits that would not directly serve the poor. The authors’ empirical results suggest that the first altruistic theory is validated by the data, and that the results matter in the sense that this altruistic difference makes a difference for the poor. Specifically, the authors were able to show that untied 76 grants made to small faith-inspired facilities were used to increase services and reduce costs for patients, as opposed to increasing benefits for staff. Yet as this example also implicitly suggests, the extent to which faith-inspired health providers are able to make their services affordable for the poor depends on their funding and other resources in comparison to those of other providers. In order to be able to provide quality services at low cost to the poor, faith-inspired providers must benefit from appropriate resources, for example through support from congregations, whether these are locally based or located in developed countries, or from other organizations including government agencies. In the absence of such support, subsidies granted to the poor may require charging better off patients more for the services provided to those groups, or relying on staffs that are willing to work at below market wages. Different strategies for reducing the cost of services for the poor may not have the same medium- or long-term consequences. For example, relying on staffs who are willing to work at below market wages (as may be the case for nuns for example), or on resources made available by external groups may not carry a risk in terms of financial sustainability as long as the staffs are willing to continue to work for low wages or as long as external funders are willing to continue to provide resources in order to make services more affordable for the poor. By contrast, differentiated subsidies for the poor paid for by asking higher fees from other groups – what could be referred to as a Robin Hood strategy - might not be sustainable under competitive markets. Indeed, under competitive markets, subsidies for the poor would lead not only to poor patients or students relying on faith-inspired facilities as compared to other facilities, but also to fewer non-poor patients or students, which would ultimately be unsustainable in the absence of other funding or cost reduction mechanisms, such as those mentioned earlier. It might be feasible under different types of markets to charge more to the better off in order to subsidize the poor – for example, under a segmented market with quality differentiated among others according to faith, better off households who value the faith affiliation of a clinic may be willing to pay more at that clinic than at another clinic, which may then make it feasible for a facility to subsidize the services provided to the poor. Yet it is not clear how much resources might be generated through price differentiation for such purposes. In this paper, we do not discuss specifically how faith-inspired facilities may be able to male their services affordable for the poor, whether this is done through cross- subsidization or through access to resources from congregations and other donors (for a case study on this question, see Gemignani et al 2012; see also more generally Wodon 2013). Instead, and more simply, we first assess in section 2 to what extent the cost of healthcare remains a major reason for not being satisfied with the health services received by households and whether concerns with costs differ between different types of providers. Thereafter, we estimate in section 3 the private cost of healthcare in a half dozen countries and compare out-of-pocket costs between public, private secular, and faith-inspired providers. Finally, we provide a brief country case study in section 4 on how health sector reforms may help in providing more equity in the financing of health facilities and reduce out-of-pocket costs for households. A brief conclusion follows. 77 IMPORTANCE OF COSTS FOR HOUSEHOLDS Is the private out-of-pocket cost of healthcare a major issue for households? In order to answer this question, we build on the analysis of patient satisfaction with faith-inspired and other service providers presented in Volume two of this series by Olivier and Wodon (2012). The results presented in that chapter suggested that satisfaction with the services provided by faith-inspired providers was higher than with those provided by public facilities. Using the same data, this section shows that one of the main reasons for non- satisfaction with the services received among households is related to the cost for households of the services. Yet at the same time, there are also substantial differences between countries, with cost being more of an issue in some of the poorest countries. The analysis is provided in tables 1 through 5 for five countries where the data are available through specific questions in the survey questionnaires. In the tables, the first line accounts for the share of households who were satisfied with the services received. The other lines represent the main reasons for not being satisfied. The responses (satisfaction and reasons for non-satisfaction) sum to 100 percent within each group (i.e., national, urban-rural, and quintiles of well-being) as well as for the various providers. In all countries, the fact that the cost of service was perceived as too expensive is the main reason for lack of satisfaction. Cost is mentioned as an issue for 37.9 percent of patients in Burundi (the poorest country in the sample), 18.0 percent in Senegal, 13.1 percent in Mali, 11.4 percent of patients in Ghana, and 10.4 percent in the Republic of Congo (the richest country in the sample). In Mali and Burundi (the two poorest countries ion the sample) but not in the other three countries, cost is also mentioned more by households in the bottom quintiles of well-being, which does makes sense. What is also interesting is the fact that in four of the five countries, cost is actually mentioned as being less of an issue for faith-inspired facilities than for public facilities. In the Republic of Congo, 14.6 percent of patients in public facilities declare that cost is an issue, versus 6.5 percent in faith-inspired facilities. In Burundi, the two corresponding figures are 37.9 percent for public facilities, versus 30.6 percent for faith-inspired facilities. In Mali, the comparison is 16.9 percent to 6.0 percent. Finally in Senegal 19.6 percent of users of public facilities complain about cost, versus only 2.9 percent in faith- inspired facilities. For Ghana by contrast, the proportion of users who complain about cost is similar in both types of facilities (it is actually slightly higher in faith-inspired facilities at 14.4 percent versus 13.2 percent in public facilities), but this is also the country where there are no substantial differences in overall satisfaction rates between public and faith-inspired providers. Note that in three of the five countries, complaints about cost were higher in other private facilities than in the faith-inspired sub-sector. The comparison with private facilities is however more complex because many households going to private facilities may have formal insurance that reduce out of pocket costs. 78 Table 1: Satisfaction rates and reasons for non-satisfaction in Burundi (%) Residence Area Welfare quintile Total Urban Rural Q1 Q2 Q3 Q4 Q5 Public None (satisfied) 46.6 37.8 36.8 35.5 37.5 35.1 44.0 38.0 Facilities were not clean 0.6* 1.6 0.6* 0.7* 0.6* 2.2 3.1 1.6 Long waiting time 6.7* 11.9 10.8 10.0 11.5 14.7 11.4 11.8 No trained professionals 2.3* 1.3 1.1* 2.0* 1.1* 1.0* 1.7* 1.4 Too expensive 35.1 38.0 41.3 40.6 37.9 39.7 31.6 37.9 No drugs available 2.2* 3.3 3.7 3.7 4.2 3.3 1.9 3.3 Treatment unsuccessful 1.1* 4.5 5.0 4.8 4.5 3.1 4.7 4.4 Prison 1.9* 0.0* 0.0* 0.0* 0.0* 0.1* 0.1* 0.1* Other 3.5* 1.6 0.6* 2.7 2.7 0.9* 1.5 1.7 Faith-inspired None (satisfied) 47.4 43.1 46.6 43.7 34.6 48.1 42.6 43.2 Facilities were not clean 0.0* 0.6* 0.0* 0.0* 0.6* 1.7* 0.5* 0.6* Long waiting time 7.1* 17.0 18.5 18.4 17.1 10.8* 18.9 16.7 No trained professionals 0.0* 0.5* 0.0* 0.0* 1.9* 0.5* 0.0* 0.5* Too expensive 38.4 30.3 29.0 30.7 39.2 32.7 21.7 30.6 No drugs available 2.7* 2.6 1.8* 3.1* 2.2* 2.8* 3.2* 2.6 Treatment unsuccessful 1.3* 4.8 2.7* 3.9* 2.5* 1.9* 12.1* 4.7 Prison - - - - - - - - Other 3.1* 1.1* 1.4* 0.4* 1.8* 1.5* 1.0* 1.2* Private secular None (satisfied) 47.7 39.3 34.9 41.8 43.3 35.9 43.2 40.0 Facilities were not clean 0.0* 2.4 1.1* 1.5* 0.8* 5.9* 0.9* 2.2 Long waiting time 2.2* 7.7 10.9* 5.4* 5.6* 8.1 6.6 7.3 No trained professionals 0.5* 1.1* 3.9* 1.0* 1.0* 0.0* 0.2* 1.0* Too expensive 41.6 42.6 39.2 47.1 42.1 42.9 42.0 42.5 No drugs available 3.4* 2.0 2.5* 1.3* 2.2* 2.6* 1.8* 2.1 Treatment unsuccessful 1.9* 3.7 5.7* 0.7* 3.5* 3.3* 4.0* 3.5 Prison - - - - - - - - Other 2.7* 1.3* 1.9* 1.1* 1.5* 1.2* 1.3* 1.4* All users None (satisfied) 47.3 38.6 37.6 37.6 38.3 36.7 43.7 39.0 Facilities were not clean 0.3* 1.7 0.7* 0.7* 0.8* 2.9 2.6 1.7 Long waiting time 4.8 11.8 11.8 10.4 10.9 13.0 11.1 11.5 No trained professionals 1.5* 1.2 1.5* 1.5* 1.2* 0.7* 1.3 1.2 Too expensive 38.1 37.9 39.4 40.3 38.9 39.5 32.6 37.9 No drugs available 2.8* 3.0 3.2 3.3 3.6 3.1 2.0 3.0 Treatment unsuccessful 1.5* 4.3 4.9 4.0 4.0 3.0 5.3 4.2 Prison 0.9* 0.0* 0.0* 0.0* 0.0* 0.1* 0.1* 0.0* Other 3.1* 1.5 1.0* 2.1 2.4 1.0* 1.4 1.6 Source: Authors’ estimations using household surveys. Note: * indicates less than 20 observations – these cells are likely not to be reliable but provided for completeness. 79 Table 2: Satisfaction rates and reasons for non-Satisfaction in Ghana, 2003 (%) Residence Area Welfare quintile Total Urban Rural Q1 Q2 Q3 Q4 Q5 Public None (satisfied) 73.4 73.2 70.6 75.7 75.0 73.0 72.1 73.3 Facilities were not clean 1.0 0.8 0.8 1.2 0.8 0.5 1.0 0.9 Long waiting time 8.3 5.7 5.8 5.6 6.2 7.9 8.0 6.8 No trained professionals 1.1 1.2 1.6 1.3 1.0 1.0 1.0 1.1 Too expensive 13.8 12.8 13.1 12.4 11.8 13.7 14.6 13.2 No drugs available 7.0 5.7 7.1 5.3 5.6 5.9 7.3 6.3 Treatment unsuccessful 5.6 8.5 10.1 6.8 6.7 5.9 7.3 7.2 Poor staffing attitude 2.5 1.9 2.0 1.8 1.7 2.3 2.7 2.2 Other 0.2 0.4 0.4 0.3 0.4 0.3 0.2 0.3 Faith-inspired None (satisfied) 73.1 72.9 67.2 76.0 74.3 73.0 74.4 72.9 Facilities were not clean 0.1 0.4 0.0 0.8 0.3 0.1 0.4 0.3 Long waiting time 6.1 6.0 8.1 3.6 5.3 3.6 9.0 6.0 No trained professionals 1.3 1.6 0.7 2.1 1.2 2.9 0.7 1.5 Too expensive 17.6 13.1 17.3 12.0 14.9 14.0 13.4 14.4 No drugs available 3.9 3.9 4.3 2.5 3.0 4.6 4.9 3.9 Treatment unsuccessful 6.3 8.4 10.6 8.5 7.1 9.2 4.0 7.9 Poor staffing attitude 1.2 2.2 2.0 2.5 3.1 1.0 1.4 1.9 Other 0.6 0.4 1.0 0.2 0.9 0.0 0.3 0.5 Private secular None (satisfied) 83.2 83.8 83.2 85.7 84.3 82.5 82.3 83.5 Facilities were not clean 0.3 0.2 0.3 0.2 0.3 0.3 0.2 0.2 Long waiting time 2.0 0.9 0.7 1.0 1.0 1.6 2.1 1.3 No trained professionals 0.4 0.3 0.3 0.2 0.2 0.5 0.4 0.3 Too expensive 11.9 8.1 7.6 7.5 9.1 11.4 11.6 9.7 No drugs available 3.6 2.6 2.5 2.0 2.6 3.9 3.6 3.0 Treatment unsuccessful 3.2 7.0 8.3 5.7 5.1 4.6 4.4 5.4 Poor staffing attitude 0.5 0.8 0.8 0.5 0.5 0.8 0.8 0.7 Other 0.2 0.2 0.2 0.1 0.1 0.3 0.3 0.2 All users None (satisfied) 78.5 78.9 77.2 81.0 79.9 78.0 77.8 78.7 Facilities were not clean 0.6 0.5 0.5 0.7 0.5 0.4 0.5 0.5 Long waiting time 4.9 3.1 3.2 3.1 3.4 4.4 4.8 3.9 No trained professionals 0.7 0.7 0.8 0.7 0.6 0.8 0.7 0.7 Too expensive 12.9 10.3 10.4 9.8 10.5 12.5 12.9 11.4 No drugs available 5.1 3.9 4.5 3.4 3.9 4.8 5.2 4.4 Treatment unsuccessful 4.3 7.7 9.2 6.3 5.9 5.3 5.6 6.3 Poor staffing attitude 1.4 1.3 1.4 1.1 1.1 1.5 1.6 1.4 Other 0.2 0.3 0.3 0.2 0.3 0.3 0.2 0.3 Source: Authors’ estimations using household surveys. Note: * indicates less than 20 observations – these cells are likely not to be reliable but provided for completeness. 80 Table 3: Satisfaction rates and reasons for non-satisfaction in Senegal (%) Residence Area Welfare quintile Total Urban Rural Q1 Q2 Q3 Q4 Q5 Public None (satisfied) 71.1 57.8 67.1 65.8 58.5 63.2 65.9 64.0 Staff not welcoming 2.1 2.9 1.5 1.1 3.6 3.2 2.5 2.5 Facilities were not clean 1.2 1.1 0.9 0.7 1.3 1.2 1.4 1.1 Long waiting time 9.2 16.3 8.0 8.8 15.7 14.9 14.7 12.9 No trained professionals 2.2 4.5 2.5 2.1 4.9 4.5 2.7 3.4 Too expensive 12.6 25.7 17.0 16.3 23.8 21.7 18.0 19.6 No drugs available 4.5 6.8 3.2 5.1 6.5 6.7 6.1 5.7 Treatment unsuccessful 3.2 9.6 8.6 8.4 6.6 6.8 4.2 6.6 Other 1.3 10.3 3.7 4.9 5.8 7.6 7.0 6.1 Faith-inspired None (satisfied) 86.9 86.6 92.8 86.2 79.3 85.9 90.9 86.8 Staff not welcoming 0.0 0.1 0.0 0.4 0.0 0.0 0.0 0.0 Facilities were not clean 2.4 0.0 0.0 0.0 0.0 0.0 8.1 1.2 Long waiting time 4.8 5.2 3.7 3.6 10.7 5.2 0.0 5.0 No trained professionals 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Too expensive 1.8 3.9 1.4 0.4 8.5 2.7 0.0 2.9 No drugs available 3.8 4.0 1.4 4.0 0.0 8.2 1.0 3.9 Treatment unsuccessful 0.0 2.9 2.4 6.9 1.5 0.0 0.0 1.5 Other 1.5 1.2 2.4 0.4 0.6 1.9 0.0 1.3 Private secular None (satisfied) 71.9 66.1 66.7 64.1 65.4 70.9 74.0 69.2 Staff not welcoming 2.9 1.0 0.6 0.7 2.0 3.6 2.4 2.0 Facilities were not clean 0.8 0.7 1.2 0.3 0.5 0.8 0.7 0.7 Long waiting time 7.1 5.1 5.1 6.6 5.4 5.4 7.3 6.1 No trained professionals 0.9 3.9 1.8 2.5 5.2 1.8 1.2 2.3 Too expensive 15.3 16.9 14.7 18.7 16.9 15.7 15.4 16.1 No drugs available 2.5 5.3 4.7 6.5 3.7 2.6 2.7 3.8 Treatment unsuccessful 4.7 10.4 11.7 9.5 6.6 5.8 5.4 7.4 Other 1.2 7.0 5.3 5.7 5.9 3.5 1.5 3.9 All users None (satisfied) 71.8 61.0 67.7 65.6 61.0 66.3 69.2 66.2 Staff not welcoming 2.3 2.2 1.2 1.0 3.0 3.2 2.4 2.3 Facilities were not clean 1.1 1.0 1.0 0.5 1.1 1.0 1.3 1.0 Long waiting time 8.3 12.6 6.9 8.0 12.6 11.8 11.8 10.5 No trained professionals 1.7 4.2 2.2 2.1 4.9 3.6 2.1 3.0 Too expensive 13.3 22.5 15.8 16.8 21.5 19.2 16.8 18.0 No drugs available 3.7 6.3 3.7 5.5 5.5 5.6 4.8 5.0 Treatment unsuccessful 3.7 9.7 9.6 8.7 6.4 6.3 4.6 6.8 Other 1.2 9.1 4.3 5.1 5.7 6.2 4.9 5.3 Source: Authors’ estimations using household surveys. Note: * indicates less than 20 observations – these cells are likely not to be reliable but provided for completeness. 81 Table 4: Satisfaction Rates and Reasons for non-Satisfaction in Rep. of Congo (%) Residence Area Welfare quintile Total Urban Rural Q1 Q2 Q3 Q4 Q5 Public None (satisfied) 68.0 62.1 69.6 68.0 60.6 63.3 67.3 65.7 Staff not welcoming 6.5 14.4 11.0 11.6 7.9 9.6 8.9 9.7 Long waiting time 13.2 19.2 14.8 11.8 22.8 17.1 11.5 15.6 No trained professionals 0.5 3.3 4.4 0.4 1.7 1.8 0.3 1.7 Too expensive 15.0 13.9 12.0 10.7 19.2 12.9 17.1 14.6 No drugs available 9.0 13.4 12.9 9.3 15.5 7.4 9.0 10.8 Treatment unsuccessful 3.9 8.6 9.8 5.9 4.7 4.9 4.4 5.8 Other 1.2 1.7 1.0 1.1 0.8 3.1 0.9 1.4 Faith-inspired None (satisfied) 89.5 91.3 80.6 100.0* 79.2* 89.4* 100.0* 90.0 Staff not welcoming 0.0 4.2 3.3 0.0* 0.0* 0.0* 0.0* 1.1 Long waiting time 3.2 0.0 0.0 0.0* 0.0* 10.6* 0.0* 2.4 No trained professionals 0.0 0.0 0.0 0.0* 0.0* 0.0* 0.0* 0.0 Too expensive 8.7 0.0 8.3 0.0* 20.8* 10.6* 0.0* 6.5 No drugs available 3.7 4.2 7.5 0.0* 20.8* 0.0* 0.0* 3.8 Treatment unsuccessful 0.0 0.0 0.0 0.0* 0.0* 0.0* 0.0* 0.0 Other 1.8 4.6 7.8 0.0* 0.0* 0.0* 0.0* 2.5 Private secular None (satisfied) 87.1 85.7 85.4 83.8 88.4 87.7 86.8 86.5 Staff not welcoming 1.7 1.6 0.5 3.7 0.8 0.3 2.9 1.7 Long waiting time 1.4 2.6 0.0 2.0 0.8 2.4 3.9 1.9 No trained professionals 0.0 0.4 0.0 0.0 0.0 0.7 0.0 0.2 Too expensive 6.9 6.9 9.1 3.8 6.9 7.5 6.8 6.9 No drugs available 2.8 2.8 3.5 5.2 3.3 1.6 1.3 2.8 Treatment unsuccessful 2.8 5.8 4.9 5.2 4.6 3.1 3.2 4.1 Other 0.6 0.7 0.9 1.8 0.0 0.4 0.3 0.6 All users None (satisfied) 78.4 75.5 78.3 77.1 74.6 77.1 78.5 77.2 Staff not welcoming 3.9 7.3 5.2 7.2 4.3 4.4 5.5 5.3 Long waiting time 7.0 9.9 6.4 6.4 11.6 9.2 7.2 8.2 No trained professionals 0.3 1.7 1.9 0.2 0.8 1.2 0.2 0.8 Too expensive 10.7 9.8 10.3 6.8 13.1 10.0 11.2 10.4 No drugs available 5.7 7.5 7.8 6.9 9.5 4.1 4.7 6.5 Treatment unsuccessful 3.2 6.9 6.7 5.3 4.6 3.8 3.7 4.7 Other 0.9 1.2 1.3 1.4 0.4 1.6 0.6 1.0 Source: Authors’ estimations using household surveys. Note: * indicates less than 20 observations – these cells are likely not to be reliable but provided for completeness. 82 Table 5: Satisfaction rates and reasons for non-Satisfaction in Mali (%) Residence Area Welfare quintile Total Urban Rural Q1 Q2 Q3 Q4 Q5 Public None (satisfied) 67.7 60.8 55.1 63.5 60.4 65.9 67.3 63.7 Facilities were not clean 0.9 1.6 2.7 2.2 0.5 0.9 1.3 1.3 Long waiting time 17.8 7.7 14.9 9.5 8.1 10.5 16.5 11.9 No trained professionals 2.3 2.2 3.1 2.0 2.5 2.8 1.1 2.2 Too expensive 9.6 22.1 24.7 22.1 23.6 13.9 8.7 16.9 No drugs available 6.1 7.7 9.0 5.1 6.4 9.2 5.7 7.0 Treatment unsuccessful 5.2 9.2 14.4 8.9 9.6 5.5 4.6 7.5 Staff not welcoming 3.3 1.1 2.2 2.0 1.5 2.3 2.0 2.0 Other 0.4 2.2 0.2 1.3 3.2 2.2 0.0 1.4 Faith-inspired None (satisfied) 45.6* 85.0 53.6* 100.0* 100.0* 85.2* 0.0* 78.7 Facilities were not clean 0.0* 0.0 0.0* 0.0* 0.0* 0.0* 0.0* 0.0 Long waiting time 9.7* 0.0 0.0* 0.0* 0.0* 0.0* 55.8* 1.5 No trained professionals 0.0* 0.0 0.0* 0.0* 0.0* 0.0* 0.0* 0.0 Too expensive 37.1* 0.0 0.0* 0.0* 0.0* 14.8* 0.0* 6.0 No drugs available 0.0* 0.0 0.0* 0.0* 0.0* 0.0* 0.0* 0.0 Treatment unsuccessful 0.0* 15.0 46.4* 0.0* 0.0* 0.0* 0.0* 12.6 Staff not welcoming 7.6* 0.0 0.0* 0.0* 0.0* 0.0* 44.2* 1.2 Other 0.0* 0.0 0.0* 0.0* 0.0* 0.0* 0.0* 0.0 Private secular None (satisfied) 75.6 78.7 84.2 82.0 76.7 66.8 76.7 77.2 Facilities were not clean 0.4 0.5 0.6 0.3 1.2 0.0 0.4 0.5 Long waiting time 15.3 4.9 2.3 2.1 7.0 19.0 15.1 9.9 No trained professionals 0.1 2.0 2.3 0.7 2.1 1.0 0.0 1.1 Too expensive 3.0 5.6 3.3 6.4 5.5 6.1 2.2 4.3 No drugs available 0.3 4.2 3.2 4.8 3.8 1.4 0.2 2.3 Treatment unsuccessful 6.5 9.8 9.9 7.9 8.2 10.7 6.0 8.2 Staff not welcoming 0.7 0.5 0.8 0.0 1.3 1.3 0.0 0.6 Other 0.0 0.2 0.0 0.0 0.7 0.0 0.0 0.1 All users None (satisfied) 70.2 66.1 67.2 69.3 65.1 66.4 70.3 67.8 Facilities were not clean 0.7 1.3 1.7 1.6 0.7 0.7 1.0 1.1 Long waiting time 17.0 6.8 9.3 7.2 7.7 12.1 16.1 11.2 No trained professionals 1.5 2.1 2.7 1.6 2.3 2.4 0.7 1.8 Too expensive 7.5 17.3 15.2 17.2 18.6 12.3 6.6 13.1 No drugs available 4.2 6.6 6.4 5.0 5.6 7.4 3.9 5.6 Treatment unsuccessful 5.6 9.4 13.2 8.5 9.1 6.5 5.0 7.8 Staff not welcoming 2.4 0.9 1.6 1.4 1.4 2.1 1.4 1.6 Other 0.3 1.6 0.1 0.9 2.5 1.7 0.0 1.0 Source: Authors’ estimations using household surveys. Note: * indicates less than 20 observations – these cells are likely not to be reliable but provided for completeness. After cost, the second main reason for non-satisfaction is long waiting time, again in virtually all countries. This was an issue for 11.5 percent of patients in Burundi, 11.2 percent in Mali, 10.5 percent in Senegal, 8.2 percent in the Republic of Congo, and 3.9 percent of patients in Ghana (in that country, the complaint ranks third after unsuccessful treatment). On this issue, FIIs do not seem to have a demonstrable comparative advantage versus public and private secular facilities. In some countries, complaints about long 83 waiting times are higher among faith-inspired facilities than among public facilities, but in other countries, the reverse is observed. As for the other reasons why some households declare being unsatisfied, sample sizes among faith-inspired facilities are often too small to be able to make a valid comparison with public facilities. What could explain the better performance of faith-inspired providers in terms of the satisfaction-cost relationship as compared to public facilities? One explanation could be that faith-inspired providers provide services at a lower cost to households – possibly through efforts to make care affordable for the poor. This is what is observed in the next chapter in this volume on the basis of qualitative work for Burkina Faso by Gemignani and Wodon (2012). But when this is the case, it must be that faith-based providers have ways to reduce their own operating costs or have access to other resources in order to be financially sustainable, given that they often benefit from smaller support from the state than is the case for public facilities. Another potential explanation often mentioned in the literature could be that patients are more satisfied with faith-inspired providers despite higher out-of-pocket costs, as discussed by Olivier and Wodon (2012). The evidence provided in the next section tends to suggest that both factors may be at work, and that the answer depends on the specific countries (and probably facilities) being considered. ESTIMATES OF OUT-OF-POCKET COSTS This section provides cross-country evidence on the cost of healthcare using multi- purpose household surveys whose questionnaires include health modules with private cost data and distinguish between public, faith-inspired, and private secular providers. These are different surveys from those used in the previous section, so that it is difficult to compare the results between on actual out-of-pocket costs and the perceptions of costs discussed earlier. Still, the data provide valuable insights on out-of-pocket costs. Average out-of-pocket costs for households of consultations are provided by type of provider in table 6 for nine countries where that information is available. These are not the total costs paid by households – for example transport costs are not included, but these are the costs paid to health facilities for the services received. The main interest is again in the comparison of the cost of public and faith-inspired facilities, but the table also provides data on other providers of healthcare. The results are somewhat unexpected. In four countries (Burundi, Cameroon, Swaziland, and Zambia), consultation costs are similar between public and faith-inspired facilities, while one might have expected costs to be substantially higher in faith-inspired facilities because they tend to benefit only from limited support from the state. This is the case in three countries (Ghana, Malawi, and Nigeria) where public facilities are cheaper than faith-inspired facilities. But in Sierra Leone care provided in faith-inspired facilities tend to be substantially cheaper for households than publicly provided care. Thus, the comparison of out-of-pocket costs tends to be country specific, and in addition, differences in costs between public and faith-inspired providers tend to be smaller on average than one might have expected given the fact that state funding for faith-inspired providers is often limited so that they need to find other ways to achieve cost recovery. 84 It should be emphasized however that these broad comparisons of costs are provided across all types of facilities, and across all types of care seeking consultations, and this may affect comparisons. For example, in Ghana most faith-inspired providers are associated with the Christian Health Association of Ghana (CHAG), and the role of hospitals in CHAG is larger in relative terms than the role of hospitals in public facilities. The fact that hospitals tend to provide more intensive care that also tends to be more expensive could explain the differences in costs observed in table 6 whereby in that country, faith-inspired providers as a whole appear to be more costly than public providers. More detailed work could try to look at differences between providers according to the type of facility used by households (say, a clinic versus a hospital), but unfortunately, this is difficult to do in most countries because the sample size tends to be too small for such disaggregation (in many countries, the market share of faith-inspired providers is small, so that slicing the observations in the sample for those seeking care by type of facility tends to reduce the sample sizes too much for estimation). Note that the same caveat applies for cost comparisons with other providers, especially because that group tends to be highly heterogeneous (it would include private secular facilities-based care as well as chemical stores and pharmacists, for example). Another issue that makes the comparison of out-of-pocket costs between providers difficult is the fact that insurances play a role – individuals who benefit from insurance and tend to be better off will typically use private secular providers more, but may have lower out-of- pocket costs because of their insurance coverage. This then may hide the true cost of care in household surveys. In the next section, a more detailed analysis of costs for Ghana will be provided with controls. But what is clear from table 6, is that as expected, faith- inspired tend to be at least as expensive as public providers, and this in turn is related to the fact that typically, faith-inspired providers receive only partial support from the state, so that they do need to request higher fees from households for cost recovery even when they benefit from a cost advantage if they can rely on dedicated staffs who may be willing to work for lower wages due to their desire to serve the poor. 85 Table 6: Cost of Healthcare Consultation, Local Currencies Residence Area Welfare quintile National Urban Rural Q1 Q2 Q3 Q4 Q5 Burundi, 2006 Public 861 230 175 160 114 118 593 244 Faith-inspired 653 192 63 41 499 58 366 208 Private secular 1558 337 137 120 351 439 879 455 Total 1164 243 155 137 203 170 622 277 Cameroon, 2007 Public 1113 492 344 486 592 773 1151 729 Faith-inspired 937 658 531 678 687 852 963 774 Private secular 841 233 96 228 321 547 960 459 Total 979 410 266 397 494 702 1052 628 Ghana, 2005/06 Public 9403 6866 4982 9567 7409 6292 9874 7902 Faith-inspired 11965 9176 4587 5469 13394 8290 14762 10343 Private secular 6600 2542 3210 1435 3510 4077 6574 4057 Total 8272 4824 4123 5207 5904 5327 8616 6175 Sierra Leone, 2003/04 Public 6266 2603 1359 1907 2541 3386 6481 4058 Faith-inspired 3529 2183 1141 1184 2552 2175 4911 2597 Private secular 7873 2208 1820 906 2479 4363 8548 5368 Total 6824 2465 1447 1515 2520 3593 7207 4407 Swaziland, 2009/10 (total medical bill) Public 71 35 22 22 29 36 84 40 Faith-inspired 76 48 18 36 36 41 101 52 Private secular 765 270 43 448 146 822 349 419 Total 312 73 22 64 50 180 172 112 Zambia, 2004 Public 5205 1841 1604 2121 2467 3347 4298 2937 Faith-inspired 7229 2667 2519 2082 2503 4292 4363 3244 Private secular 8214 3379 2602 3136 4771 3891 8893 5242 Total 6584 2459 2010 2513 3291 3620 6269 3847 Malawi, 2004 Public 109 18 10 9 20 27 68 27 Faith-inspired 714 226 110 153 177 217 410 244 Private secular 154 26 11 17 22 44 74 36 Total 144 31 13 18 27 46 90 40 Nigeria 2003/04 Public 529 506 185 191 284 401 937 516 Faith-inspired 1608 575 163 232 167 941 1478 997 Private secular 676 601 124 303 332 521 979 633 Total 621 553 156 244 303 464 973 582 Source: Authors’ estimations using household surveys. REDUCING COSTS FOR HOUSEHOLDS: A BRIEF CASE STUDY It was mentioned in the previous section that Ghana is one of the countries in table 6 where out-of-pocket costs for faith-inspired providers tend to be higher than for public providers. In this section, more details are provided on Ghana’s health system to suggest why this may have been the case at the time the household survey data used in table 6 86 were collected (in 2005-06 for that country), but also why this may be changing due to efforts in the national health policy to reduce out-of-pocket costs for households. At independence, Ghana inherited a publicly funded health system that comprised mostly of government-owned health care service delivery. User fees were abolished making health care theoretically free. But facing economic contractions and shortfalls in funding for the health system, public health care facilities began charging user fees for services, as did faith-inspired facilities which in the absence of public subsidies could also rely on donations from foreign missions. While cost recovery was justified as a means to encourage greater accountability of health care providers while increasing facility-level autonomy over use of internally generated funds, user fees created barriers to accessing health care in the absence of appropriate risk pooling mechanisms in place. Concurrent with the formalization of user fees, a policy to exempt the poor accompanied the system, at least for public facilities. But these exemptions failed to reach their target groups. Looking at health facilities in the Volta region, Nyonator and Kutzin (1999) found that exemptions were applied unevenly across health care facilities and income groups. They concluded that despite the policy to exempt ‘paupers’, the poor were either facing catastrophic health expenditures or were not seeking health care when needed. As a result, the poor continued to pay out-of-pocket for user fees at public (and faith-inspired) facilities. Out-of-pocket payments, the most regressive form of financing were the major source of funds for all facilities, whether public, faith-inspired, or private non-religious. At the time of the last National Health Accounts in 2002, out-of-pocket payments accounted for about half of total health expenditures. Under these circumstances the evidence on whether faith-inspired facilities were able to subsidize services for the poor and be less costly than other facilities was mixed. For example, Asenso-Okyere (1995) suggested that “since the quality of service is high, users are willing to pay for it. The poor who cannot pay the fees are exempted. Exemption can pose identification problems but since the missions operate mostly in the rural areas where people know each other it is usually not too difficult for a social worker to recognize the needy.� By contrast Nyonator and Kutzin (1999) suggested that facilities “insist on the payment of deposits…for inpatient services, especially in mission hospitals. While it is fair to say that deposits help to promote cost recovery, they also pose a serious threat to accessibility.� There have been however important changes in recent years in health service delivery in Ghana, so that what may have been observed in the past is not necessarily still relevant today. First, in comparison with other countries, CHAG has experienced an active collaborative relationship with the Ghanaian government. A Memorandum of Understanding was signed in 2003 between CHAG and the Ministry of Health, which has helped provide more state funding to CHAG facilities, especially for salaries. As of 2005 the government provided between 45 percent and 60 percent of CHAG’s total operating revenues (CHAG 2006), and this may have increased further since. About 80 percent of this was for the salaries of about 7,000 CHAG health staffs. Thus in contrast to the time when CHAG facilities operated as mission organizations, they now obtain most of their revenues from government funding for salary costs and from the services they provide. 87 In addition, in an effort to remove user fees, provide protection against catastrophic health expenditures, and improve equity and access to care, Ghana created in 2003 the National Health Insurance Scheme (NHIS). The NHIS relies on a combination of earmarked tax revenues and income-based premium contributions to substitute for user fees. In this system, the funding follows the patient. Health care providers are reimbursed by the NHIS for services and drugs covered in the benefits package that are provided to NHIS members. The NHIS exempts vulnerable groups such as the indigent, the elderly, and children of enrolled members from paying premiums. The premium contribution levels are also income-based in order to reduce the burden on the poor. While the legislative instrument guiding the NHIS did set a narrow definition of the indigent category as a group with no identifiable source of income or shelter and fixed the percentage of enrolled members that qualify as indigent at 2.5 percent of those enrolled (versus a share of the population in poverty of 28.6 percent), the policy still has helped. According to data from the NHIS website, as of June 2009 some 13.8 million individuals were registered in the scheme, representing close to 60 percent of the population. The program is managed by District Mutual Health Schemes, but in addition to members participating through district schemes, workers contributing to the Social Security and National Insurance Trust are also enrolled. On the provider side, public facilities are automatically accredited to participate, but in addition the NHIS has provisionally accredited 1,551 private healthcare facilities, including 400 hospitals and clinics, 237 maternity homes, 451 pharmacies, 329 licensed chemical shops and 128 diagnostic facilities (laboratories and diagnostic imaging facilities). Most of the CHAG facilities have been accredited. Makinen et al (2011) assess that CHAG currently has the highest share of NHIS-covered patients (73 percent), compared with public providers (69 percent) and self-financed private providers (56 percent). Overall, exempt groups, which do not have to pay to be part of the scheme, accounted for close to 70 percent of all registered members. By contrast, the estimations presented in table 6 were obtained with the GLSS5, which dates back to 2005-06, a time at which only a small share of the population has registered with the NHIS 19. Thus the situation has probably improved. According to CHAG management, the NHIS has facilitated payments to CHAG member facilities for the services they provide and this is perceived by these facilities as a major improvement which may have further aligned the cost of care among FIIs and public facilities. Some issues may remain. For example, Ballou-Aares et al (2008) note that the rapid scale up of the NHIS has put a burden on claims processing. Administrators of CHAG facilities interviewed by Shojo et al (2012) cite delays in receiving funds, which affects their cash flow. As the Director of a CHAG hospital explained: “the idea of the NHIS is perfect. It is good for the poor and brings clinics to a certain standard. But delivery has some problem. Our workload increased. It put stress on our finance because payment does not come regularly.� Shojo et al (2012) also suggest that FIIs continue to 19 Prior to the NHIS, some mission hospitals had actually instituted their own insurance schemes (e.g., the community insurance plan around St. Theresa’s Hospital in the Brong Ahafo Region) and this experience helped inform the introduction of the NHIS. 88 suffer from a lack of resources including medicines and technical equipment, with the situation more severe for facilities not yet accredited with the NHIS. Nevertheless, the need to rely heavily on out-of-pocket costs from patients to achieve cost recovery in faith-inspired facilities has greatly been reduced with the introduction of the NHIS and the memorandum of understanding signed between CHAG and the Ministry of Health, which shows how health reforms may work towards achieving more equity in financing. CONCLUSION Faith-inspired providers often aim to serve all – and many also have a commitment to serve the poor and vulnerable. However, the extent to which they are actually able to do so (given their resources) remains an open question. Even if FIIs do benefit from staff that are dedicated, some of whom may be able and willing to work for very low pay, running a health facility does cost money, and financial sustainability requires funding. When faith-inspired facilities do not benefit from state support, or when they benefit from lower levels of state support than public facilities, they often need to rely on cost recovery from patients and students to break even. When a higher level of cost recovery is required from users, it is more difficult for faith-inspired facilities to serve the poor, because the cost of their services becomes less affordable for those in need. In this context, two questions were asked in this paper. First, is the cost of healthcare a major reason for not being satisfied with health services and does this differ between types of providers? Second, what is the out-of-pocket cost of healthcare for households? Overall, the results suggest that cost does remain a major concern for many households and that there are some differences in out-of-pocket costs between providers. Yet at the same time, these cost differences tend not to be as large as one might have assumed, and health sector reforms can help in reducing both these differences and out-of-pocket costs. REFERENCES Asenso-Okyere, W. K. 1995. “Financing Healthcare in Ghana.� World Health Forum 161:86-91. Ballou-Aares, D., A. Freitas, L. Kopczak, S. Kraiselburd, M. Laverty, E. Macharia, and P. Yadav. 2008. Private Sector Role in Health Supply Chains: Review of the Role and Potential forPrivate Sector Engagement in Developing Country Health Supply Chains. New York: The Rockefeller Foundation, Dalberg, MIT Zaragoza. Christian Health Association of Ghana (CHAG). 2006. Annual Report: June 2005 - May 2006. Accra: Christian Health Association of Ghana. Donabedian, A. 1992. “Quality Assurance in Health care: Consumers’ Role.� Quality in Health Care 1: 247-251. 89 Hadorn, D. 1991. “The Role of Public Values in Setting Health Care Priorities.� Social Science & Medicine 32: 773-781. Gemignani, R, C. Tsimpo, and Q. Wodon. 2012. “Making Quality Care Affordable for the Poor: Faith-Inspired Health Facilities in Burkina Faso.� In J. Olivier and Q. Wodon, eds, Mapping, Cost, and Reach to the Poor of Faith-Inspired Health Care Providers in sub- Saharan Africa. World Bank, Washington, DC. Makinen, M., S. Sealy, R. A. Bitran, S. Adjei, and R. Munoz. 2011. Private Health Sector Assessment in Ghana. World Bank, Washington, DC. Marek, Y., C. O’Farrell, C. Yamototo, and I. Zable. 2005. Trends and Opportunities in Public-Private Partnerships to Improve Health Service Delivery in Africa. World Bank, Washington DC. Ministry of Health. 2007. National Health Policy: Creating Wealth Through Health. Accra: Ministry of Health, Ghana. Nyonator, F., and J. Kutzin. 1999. “Health for Some? The Effects of User Fees in the Volta Region of Ghana.� Health Policy and Planning 14 (4):329-41. Reinikka, R., and J. Svensson. 2010. “Working for God? Evidence from a Change in Financing of Not-for-Profit Healthcare Providers in Uganda.� Journal of the European Economic Association 8:1159-78. Salisu, A., and V. Prinz. 2009. Healthcare in Ghana. Wien: Austrian Centre for Country of Origin and Asylum Research and Documentation, and Austrian Red Cross. Shojo, M., C. Tsimpo, and Q. Wodon. 2012. “Satisfaction with and Reasons for Choosing Faith-inspired health care provision in Ghana.� In J. Olivier and Q. Wodon, eds, The Comparative Nature of Faith-inspired Health Care Providers in sub-Saharan Africa, World Bank, Washington, DC. Wodon, Q. 2013. Faith and Human Development: Education and Health Services in Africa, New York: Palgrave MacMillan. 90 CHAPTER 6 COMMUNITY HEALTH ASSETS MAPPING: A MIXED METHOD APPROACH IN NAIROBI John Blevins, Sandra Thurman, Mimi Kiser, and Laura Beres Interfaith Health Program, Emory University This paper summarizes an innovative approach to health asset mapping in Mukuru, an informal settlement in Nairobi, Kenya. This particular approach combines a model for community based participatory research (CBPR) with a GIS mapping process. The essay provides background on the rationale for the mapping and of the characteristics of the Mukuru community, a summary of findings and insights gained from the mapping, and a description of the possibilities and challenges of a proposed initiative to mobilize the assets identified as part of a structural HIV prevention initiative. INTRODUCTION In 2008, the Interfaith Health Program (IHP) of the Rollins School of Public Health at Emory University received funding through the Kenya Office of the President’s Emergency Plan for AIDS Relief (PEPFAR) to employ a mapping process in Mukuru, a large informal settlement in Nairobi, Kenya with an estimated population ranging from 200,000 (Driscoll et al 2006, Langford et al 2006) to over 600,000 people (CDT 2007). The community health assets mapping activities were part of the first phase of an innovative HIV prevention initiative called Partnership for an HIV Free Generation (HFG). The HFG project consisted of a series of targeted social and health messaging activities designed to address the broader social drivers that are known to impact risk of HIV infection and likelihood of disease progression; Mukuru was chosen for HFG activities for two reasons. First, while a robust body of research demonstrates the pervasiveness of these social drivers in informal settlements, less is known about which activities impact them to reduce health disparities. Second, Mukuru was chosen instead of other informal settlements (such as Kibera, for example) because there were relatively fewer programs in place in there. Although there were reasons for choosing Mukuru numerous challenges also remained. Mukuru had little or no infrastructure investment and epidemiological data on HIV prevalence or incidence did not exist. Furthermore, HFG program staff had little knowledge of NGOs and CBOs operating in Mukuru. IHP was funded to conduct the mapping to identify the institutions and resources already at work and to understand how the community related to them. The methodology combines community based participatory research (CBPR) with GIS mapping to identify the intrinsic but often unknown health assets that are found in every community. 91 MUKURU IN CONTEXT Mukuru is a network of villages comprising a single catchment area in Nairobi located along Mombasa Road between the city center and Kenyatta airport, just off of Airport North Road. As an informal residential settlement, Mukuru is characterized by high poverty, semi-permanent residential structures (most structures are built of corrugated metal and large sections of Mukuru have been bulldozed in the past), little infrastructure, and few social support or health services (Gulyani et al 2010a). Most residents of Mukuru have migrated to Nairobi from rural areas in an effort to find employment and economic opportunities. Unable to secure affordable housing elsewhere, the residents eventually settle in Mukuru. The rate of chronic disease, diarrheal diseases, and infectious diseases is high with extant research identifying the primary drivers of infection to be overcrowding, an almost complete absence of drainage and sanitary systems, and little access to municipal water (Gulis et al 2004). Because it is located near heavy industries, Mukuru is often a first home for Kenyans migrating from rural areas because of low rents and close proximity to employment (Gulis et al 2004). There are no paved roads, proper garbage collection equipment, or solid waste disposal facilities. There is no storm water drainage and in rainy seasons some structures are destroyed by floods. Liquid waste is deposited in open channels that line dirt roads and residents have no access to sanitation services (Gulis et al 2004). Mukuru is not part of the existing electrical grid although illegal power lines are routinely spliced from the existing power grid with exposed connections. Most people use kerosene, charcoal, or firewood for cooking and refrigeration is rare (Langford et al 2006, CDT 2007). THE PARALLEL MAPPING METHODOLOGY In a context such as Mukuru in which infrastructure is lacking and data on socio- economic and health indicators are minimal, governmental and non-governmental organizations face tremendous challenges in establishing and maintaining comprehensive community level HIV prevention and referral initiatives such as the one proposed under this project through HFG. Efforts to address social drivers of risk and to establish community-level interventions represent a more recent focus of sustained HIV prevention efforts with behavioral researchers agreeing that such interventions are integral to comprehensive HIV prevention but endeavoring to ascertain which are most effective (Halperin et al 2004, Horton and Das 2008, Piot et al 2009, Wilson and Halperin 2009, Merson et al 2008, Coates et al 2008, Auerbach et al 2009). Despite the fact that this research is ongoing and emerging, there is strong consensus that such efforts can only be successful if the community and its assets are understood. This approach to mapping provides such understanding. The method, Participant Inquiry into Community Health Assets (PICHA 20), entailed comprehensive community-level 20 In Kiswahili, picha means “snapshot� or “picture.� The acronym of PICHA was chosen to highlight the ways in which the approach provides the community and researchers with such as snapshot of the community in which the PICHA methodology is employed. 92 workshops on the health assets of the community combined with geospatial mapping of those assets utilizing Global Information Systems (GIS) technology. Community health assets mapping is an established approach to community based participatory research and action that draws on the formative work of Robert Chambers and his colleagues. Various researchers and practitioners have drawn on Chambers’ formative work to develop mapping methodologies tailored to specific contexts. PICHA was developed out of IHP’s earlier research in collaboration with the African Religious Health Assets Programme (ARHAP), specifically a workshop model known as Participant Inquiry into Religious Health Assets, Networks, and Agency (PIRHANA) that used Chambers’ work as a foundation. IHP and ARHAP received funding from the World Health Organization (WHO) in 2005 to employ PIRHANA in Lesotho and Zambia and to combine this participatory method with GIS mapping of newly identified assets (ARHAP 2006). Community health asset mapping assumes that (1) all communities possess intrinsic assets and resources regardless of their social-economic status; (2) those resources are often under-appreciated by outside entities such as because they do not align with the structural emphases of these entities; (3) despite this misalignment, community members and macro-level institutions share a common commitment to reducing health disparities and providing opportunities for community development; and (4) in order to accomplish these common commitments, programs must learn to collaborate and build trusting relationships with the health assets and resources of local communities by following their cultural norms, values, and patterns. PICHA included two days of interactive workshops with a purposive sample of community members and community members from the villages of Mukuru kwa Njenga, Mukuru kwa Reuben, and Lunga Lunga. Modules in these workshops included: ⋅ Community History: Community Assets: Why are they valued as an asset? What qualities or characteristics does an asset possess? (e.g., it is trusted, it works to build partnerships both within community members and other organizations) ⋅ What factors contribute to health in the community? ⋅ What factors contribute to illness? ⋅ How do the identified assets contribute to health and mitigate against illness? ⋅ What are the exemplar organizations in Mukuru that have earned the community’s trust and that offer significant and sustained activities in the community? ⋅ What are the community’s priorities? ⋅ How do those priorities align with the priorities of HFG? RESULTS Prior to the workshops HFG identified a total of 54 organizations providing services in the sectors of health, education, social support, and economic empowerment/finance in the Mukuru catchment area. Workshop participants in 2008-2009 identified 194 community health assets in only three of the seventeen villages that comprise the Mukuru catchment area. In 2010, the 194 organizations were surveyed to gather information on the scope of services offered and the numbers of people served. In that process, an additional 78 organizations were identified in the three villages, bringing the total of community health assets named to 272, or an increase of over 500 percent in the number 93 of known community health assets when compared to the initial census of 54. This increase was seen after mapping only a fraction of the entire catchment area. Organizations are in flux but turnover is not as high as had been anticipated. Of the 272 total health assets identified between 2008 and 2010, 15 (all from the initial 194) were no longer operating in 2010; 94.5 percent of health assets identified were still in operation. By August 2010, 257 of the 272 community health assets had been surveyed and the GIS waypoint data for their geospatial locations had been recorded. The GIS data literally puts these intrinsic community health assets “on the map� for the first time, allowing these assets to become part of broader networks. Insights gained from parallel mapping The Mukuru project has been only one of a series of ongoing health assets mapping initiatives undertaken by IHP at Emory. These initiatives have been identified as a best practice by US-AID and PEPFAR (Jaskiewicz et al 2009) and have been employed in southern Africa by the World Health Organization (ARHAP 2006). The findings from this broader body of applied research from IHP and from similar initiatives by others (Chirowodza et al 2009) demonstrate the importance of this parallel mapping methodology in a variety of contexts. The process yields many insights specific to informal communities. Because informal settlements are generally not recognized by national governments, few resources specific to building infrastructure are devoted to these areas. The network of formal recognized roads that are common on GIS programs are missing for these settlements and the geographic space of informal settlements often appears “blank� on GIS maps. Because no pre-existing maps of the settlements exist, it is important not only to gather data on the waypoints (latitude and longitude of specific organizations) but also to geospatially plot the roads and paths of the settlements. Many of the community health assets in informal settlements do not have a physical “home.� They exist and provide services within the residences of members of the community. For example, in Mukuru 23 percent of community health assets that were identified were social support groups for women and young people operating without a facility. Most of these organizations were in the social service sector (83 percent of all social service organizations in Mukuru were providing services without dedicated office space). If macro-level funders ignore these groups, they will miss a large percentage of the community assets. These findings parallel those from a survey of microenterprise initiatives in Nairobi’s slums (Gulyani et al 2010b). That survey revealed that household- based microenterprise initiatives are common in Nairobi’s informal settlements. Although they provide employment and measurable economic benefits, they are largely invisible outside of the community in which they operate. Religious organizations provide the majority of formal organizational structure in informal settlements because they have the capacity to provide community-wide services and the physical space to do so. In Mukuru, faith based organizations comprise 83 percent of all community based organizations and 35 percent of all educational programs. Though FBOs comprise only 9 percent of the health organizations identified in the 94 mapping, that 9 percent provides 29 percent of all patient encounters and 100 percent of the intensive outpatient or inpatient medical services in Mukuru. Other insights from PICHA workshops parallel those gleaned through IHP’s other mapping activities. The parallel mapping approach provides knowledge not only of topographical features but also of the ways in which communities are affected by those features. For example, the mapping workshops in Mukuru revealed that a plot of land being considered by the CDC as the future site for a clinic would be inaccessible during heavy rains because of flooding of a nearby river; this knowledge led to a change in location of the clinic. Further, because the parallel approach is grounded in mutual, respectful relationships with local communities, it contributes to trust between members of those communities and outside entities employing the process who value communities’ wisdom, knowledge, and intrinsic resources. Adding GIS mapping to existing participatory mapping approaches addresses some earlier shortcomings of participatory community mapping alone, namely the lack of spatial accuracy and the common problem of imprecise scale (Gould 1973, Tversky 1993). Despite the benefits of GIS mapping, some proponents of participatory community mapping have been wary of using it, arguing that it is used only by outside entities to give them specialized, technical information that is of little or no use to the community-at-large and that is not then shared back with the community (Abbot et al 1998). IHP employs open source data management processes (Open Street Map) so that the data can be accessed, edited, and utilized by anyone with access to internet connectivity. We also make the data generated through GIS processes available to communities in the form of paper directories and maps. 95 Figure 1: Mukuru kwa Ruben: One Village, Four Maps Figure 1: Street map view of Mukuru kwa Ruben from Figure 2: Satellite view of Mukuru kwa Ruben from Google Maps Google Maps Figure 3: Map of Mukuru kwa Ruben drawn by Figure 4: Internet-based map of Mukuru kwa Ruben PICHA participants on Open Street Map with community health assets identified The village of Mukuru kwa Ruben is represented by the four maps above to demonstrate the process of data gathering through community health assets mapping. Figure 1 is a screen shot of the Google street map of the Mukuru kwa Ruben catchment area, which is depicted as low density, suburban-scale. Figure 2 is a satellite view of the same area reveals the actual density of kwa Ruben, which is typical of informal settlements. Figure 3 shows the hand-drawn map created by PICHA workshop participants in kwa Ruben with 41 community health assets identified [Note: Participants drew this map “upside down� in relation to the GIS maps]. Figure 4 is a map that can only be created as a result of the PICHA workshop. In it, the previously unknown community health assets were entered into the Open Street Map platform (http://www.openstreetmap.org). 96 Moving beyond mapping toward mobilization IHP’s parallel mapping activities comprise the first phase of a multi-phase initiative in Mukuru. Having identified a broad spectrum of previously-unknown community health assets, the initiative is now moving into a second phase of mobilizing those identified assets by aligning exemplar organizations and smaller, less formal community health assets by providing funding for collaborative activities to be developed to address social drivers of health disparities in Mukuru. Phase One: Mapping Assets Phase Two: Mobilizing Phase Three: Moving to (2008-2009) Assets (2009-2013) Sustainability (2009-2013) • Activities • Activities • Activities • Community Health • Establish Community • M/E reports of the Assets Mapping Advisory Board and community-level • GIS Mapping Youth Advisory Board initiatives presented to • Establish community HFG staff and • Outcomes community. priorities, align with • 272 community health HFG priorities, and • Determine which assets identified develop M/E plan. initiatives have the • 11 "exemplar • Develop ca. 20 greatest impact and are organizations" identified community-level the most viable as long- • HIV-specific services of initiatives that address term programs. the 272 community established priorities to • Develop a sustainability health assets identified be implemented and transition workplan • Profiles of the exemplar through the health for those initiatives, organizations assets with exemplar identifying scope and completed organizations taking a nature of community • Published resource lead. involvement, long-tern directory identifiying the • Where feasible, build goals, and funding 272 community health capacity of exemplar sources. assets organizations and commnuty health assets to encourage long-term sustainability IHP envisions a model of community mobilization in Mukuru that employs a combination HIV prevention approach. Combination HIV prevention methods assume that singular prevention initiatives will be insufficient to effect the myriad, complex, inter-related behavioral, environmental, and psycho-social factors that contribute to HIV risk. Only by developing a series of distinct but overlapping prevention programs - combination initiatives - can a comprehensive community HIV prevention plan be developed. In developing such an initiative in Mukuru, IHP faculty are employing research in structural prevention efforts (Gupta 2008). Structural prevention works to identify and address the various social determinants that contribute to health disparity and disease. As an emerging field of research, central questions are emerging in relation to structural prevention approaches. Public health interventions have tended to be formulated at the macro level (the level of public policy - a level that is too broad to account for differing community needs, patterns, and norms) or at the micro level (the level of individual behavior change - a level too narrow to account for the reality that people are part of various social networks and their behavior changes as they traverse these networks). Rarely are interventions formulated at the community level, with the community in mind. What would the characteristics of community-level interventions be? Until recently, most prevention 97 initiatives have conceived of human beings as separate, discreet individuals. Certain schools of social science conceive of human beings as parts of multiple, contradictory, simultaneous social networks. In this conception, human beings traverse various social locations and those social locations provide different messages and articulate different norms in regard to health behaviors. The Mukuru project aims to apply this latter conception in developing the kinds of programs that can contribute to a community’s health. What kinds of conceptual frameworks help us to understand this complexity? The challenges of public health programming at the community level with people who are traversing a number of divergent social contexts within that community is daunting. IHP proposes a four-tiered mixed method approach to monitoring and evaluation of the mobilization initiative in an effort to address those challenges. 1. A baseline of incidence/prevalence: Because there are no existing epidemiological sources to provide data on HIV incidence and prevalence in Mukuru, we propose a triangulation approach drawing from multiple sources: ⋅ Household survey by Ministry of Health: HFG staff described an existing household survey that was completed recently for the Mukuru catchment area. ⋅ Data from VCT sites in Mukuru: VCT sites in Mukuru can provide data on number of people tested and on numbers of positive tests. ⋅ Data from health clinics: We propose working with two primary care clinics—a CDC clinic and the Mukuru Promotion Centre Clinic/Mary Immaculate Clinic—in order to conduct a randomized chart review to measure HIV prevalence. ⋅ Kenyan AIDS Survey data: Data gleaned from these sources would be compared with district-level data from the Kenyan government. 2. Description of social drivers for HIV risk: In order to gain knowledge regarding the kinds of social factors that contribute to HIV risk, we propose conducting key informant interviews and in-depth narrative ethnographies of a select group of residents. While such an approach is limited in regard to generalizability, it will offer rich data for the specific context; such data are important for understanding the particular intersection of myriad social factors as they play out in Mukuru. 3. Outcome/Impact evaluation of projects: As the community develops the projects funded in this mobilization phase, IHP will gather baseline, midpoint, and endpoint data from participants in each project in order to assess changes in knowledge, skills, and attitudes pertinent to negotiating HIV risk as a result of each project. 4. Assessment of capacity-building for exemplar organizations: As Phase II of the project works to build capacity and sustainability for the most promising projects, exemplar organizations will play a key role. IHP will survey those organizations to measure the scope of services currently offered, interview staff to identify common challenges and barriers, and provide administrative oversight to the funded projects. Methodological questions about the possibility of measuring the effects of these kinds of initiatives in relation to HIV incidence and prevalence remain. Is the patient population in 98 the two clinics representative of prevalence in all of Mukuru? For example, some data show that Muslim residents of Lunga Lunga, for example, do not readily access care at Mary Immaculate Clinic, the largest of the primary care clinics, because they perceive a tension between their religious beliefs and the religious foundations of the clinic. If, at the end of the project, or at preset evaluation points, a clinic survey reveals an increase in the number of HIV-positive patients seeking treatment can we attribute this to improved transition into care? Is an increase in prevalence a negative outcome (new numbers mean prevention efforts are ineffective) or a positive outcome (project efforts improved a system of referral into treatment networks)? We also propose establishing a baseline on HIV incidence by analyzing data from VCT sites in Mukuru. Again, challenges arise in regard to this approach. Would changes in numbers of newly reported cases be due to an actual change in incidence, to improved VCT efforts, to a lowering of stigma, or something else? Is data on HIV-positive patients representative of all HIV-positive residents (in other words, what percentage of HIV-positive patients are unaware of their status and reluctant to be tested)? This question is not new to the Mukuru project but it does confound efforts at monitoring and evaluation. If the community decided through its mobilization efforts to increase VCT opportunities and those sites were successful in testing large numbers of residents, the resulting discovery of HIV-positive residents—a positive outcome in regard to the specific initiative of increased VCT testing—could result in an increase in reported cases, leading to a perceived rise in incidence. If prevalence or incidence are measured as singular, sufficient markers without further consideration of such inter-related factors then there would be an inherent, implicit disincentive in the project to identify those infected because it would bias a data analysis of the project toward a conclusion that the project actually contributed to incidence. CONCLUSION The parallel mapping methodology employed by IHP in Mukuru is a powerful tool for mapping the health assets of local communities. It is a resource for gaining information about organizations in local communities that might not be known to outside entities. In the Mukuru initiative, the mapping identified 272 previously unknown community health assets, an increase of over 500 percent from the number known prior to the initiative. In addition, the method provides key qualitative information not only about the raw numbers of assets but also about the ways in which they relate to the community and the level of trust within the community regarding those assets. The method elicits from community members their own priorities and provides ways to align community priorities with those of outside organizations. Finally, it provides a process for standardizing information and location of community health assets through comprehensive surveys and GIS technologies. This kind of knowledge is imperative for public health initiatives involving community-level interventions and for systems capacity assessment processes that are part of health systems strengthening. Further, this methodology respects local communities and the people who make up those communities. This respect is the first, critical step in developing collaborations between 99 communities and governmental and non-governmental organizations, the kind of collaboration that has been far too uncommon in public health programs up to this point. REFERENCES Abbot, J., et al. 1998. “Participatory GIS: Opportunity or Oxymoron?.� PLA Notes 33: 27–33. African Health Assets Programme (ARHAP). 2006. Appreciating Assets: The Contribution of Religion to Universal Access in Africa. Cape Town: African Religious Health Assets Programme. Auerbach, J., et al. 2009. “Addressing Social Drivers of HIV/AIDS: Some Conceptual, Methodological, and Evidentiary Considerations.� aids2031 Social Drivers Working Group. Chirowodza, A., et al. 2009. “Using Participatory Methods and Geographic Information Systems (GIS) to Prepare for an HIV Community-Based Trial in Vulindela, South Africa.� Journal of Community Psychology 37(1): 41-57. Coates T, et al. 2008. “Behavioural Strategies to Reduce HIV Transmission: How to Make them Work Better.� The Lancet 372: 669-684. Community Development Trust Fund. 2007. “Improving Sanitary Conditions for Slum Dwellers in Nairobi, Kenya.� A Collaborative Project with the European Commission, Nairobi. Driscoll, D., et al. 2006. Planning for Community-led Upgrading in Mukuru kwa, Mimeo. Gould, P. 1973. “On Mental Maps.� In R. Downs and D. Stea, eds, Image and Environment, London: Edward Arnold. Gulis, G., et al. 2004. “Health Status of People of Slums in Nairobi, Kenya.� Environmental Research 96: 219–227. Gulyani, S., et al. 2010a. “Poverty, Living Conditions, and Infrastructure Access: A Comparison of Slums in Dakar, Johannesburg, and Nairobi.� World Bank, Washington DC. _____. 2010b. “Inside Informality: The Links Between Poverty, Microenterprises, and Living Conditions in Nairobi’s Slums.� World Development 38 (12): 1720-1726. Gupta, G. R., et al. 2008. “Structural Approaches to HIV Prevention.� The Lancet 372: 764-75. 100 Halperin, D., et al. 2004. “The Time has Come for Common Ground on Preventing HIV Transmission.� The Lancet 364: 1913-1915. Horton, R., and P. Das. 2008. “Putting Prevention at the Forefront of HIV/AIDS.� The Lancet 372: 421-422. Jaskiewicz, W., L. Fitzgerald, L. Fogarty, A. Huber, G. Peersman, S. Schalk-Zaitzev, R. Sethi, M. Tholandi, and S. Yank. 2009. Promising Practices to Build Human Resources Capacity in HIV Strategic Information. Chapel Hill, NC: The Capacity Project. Langford, M., et al. 2006. “Listening to the Poor? Housing Rights in Nairobi, Kenya.� Geneva: Centre on Housing Rights and Evictions. Merson, M., et al. 2008. “The History and Challenge of HIV Prevention.� The Lancet 372: 475-488. Piot, P., et al. 2009. “AIDS: Lessons Learnt and Myths Dispelled.� The Lancet 374: 260- 263. Tversky, B. 1993. “Cognitive Maps, Cognitive Collages, and Spatial Mental Models.� Lecture Notes in Computer Science 716:14–24, Berlin: Springer. Wilson, D., and D. Halperin. 2009. “`Know Your Epidemic, Know Your Response’: A Useful Approach if We Get it Right.� The Lancet 372: 323-326. 101 CHAPTER 7 MAKING QUALITY CARE AFFORDABLE FOR THE POOR: FAITH- INSPIRED HEALTH FACILITIES IN BURKINA FASO Regina Gemignani, Clarence Tsimpo and Quentin Wodon 21 The World Bank Based on the results of qualitative fieldwork conducted in 2010 in Burkina Faso, this paper suggests that a key reason for individuals to seek care in faith-inspired health facilities is the fact that the cost of care is lower than in public facilities (the other reason being that faith-inspired facilities appear to provide better quality of care). Yet faith-inspired facilities receive only limited support from the state to provide their services. The ability of the facilities to make quality care affordable for the poor is maintained thanks to support in kind and in cash from religious groups and other donors. This model contributes substantially to the availability of affordable quality care in the communities where the facilities are located, but higher state support for the facilities would help for expansion. INTRODUCTION Faith-inspired institutions (FIIs) play an important role in improving health care in many African countries through the delivery of facility-based care, training of the health workforce, involvement in community-based activities, and participation in health promotion campaigns. FIIs represent a wide array of faiths and diverse motivations and goals. While a number of factors shape the approaches taken by FIIs in planning their activities, an often expressed priority is to serve the poor for whom care often remains unaffordable. Many faith-based providers also aim to provide preventive and/or curative care to those who are at a disadvantage or marginalized (see for example Wodon 2013). In the case of Burkina Faso, significant progress has been achieved to facilitate access to care, among others the construction of new facilities, vaccination campaigns, and the reduction or elimination of selected fees. As a result, many health statistics such as the rate of assisted childbirth, the use of health services, vaccination, and infant mortality have improved (WHO 2009, USAID 2009). Nevertheless, available household survey data suggest that apart from cases when an individual does not seek care because there is no need for it, or self-medicates when ill or sick, the main reason for not seeking care is cost. Our estimates presented in this paper suggest that care remains too expensive, and is cited in national surveys by one third of those not seeking care when in need. Pokhrel (2010) also provides empirical evidence on the barrier to care that cost represents using 21 This work benefitted from funding from the World Bank, with special thanks to Ishac Diwan and Rakesh Nangia for supporting this research. The authors are also grateful to Jill Olivier for many enlightening discussions as well as contributions to the literature review. Boubacar Cisse and Sidi Barry collected the qualitative data and their insights from the field provided depth and fuller understanding to the analysis. 102 data from Nouna Health District, confirming previous results obtained by Baltussen et al (2002), suggesting that high costs of healthcare play a central role in lowering patient satisfaction with services. In addition, due to gender inequities in access to household resources, women are often at a disadvantage in respect to healthcare utilization (Faye 2008, Hampshire 2009). Nikiema et al (2007) describe the burden on women in Burkina Faso who are faced with negotiating access to healthcare despite their limited bargaining power. To the extent that FIIs pay special attention to making care affordable for the poor, they may help in facilitating access to care, including for women. There are comparatively fewer FIIs involved in health care in Burkina Faso than in other countries (WHO identified 60 faith-inspired facilities in the country), but these facilities may play a very important role in the communities they serve. Based on the results of qualitative fieldwork conducted in six facilities in 2010, this paper suggests that indeed a key reason for individuals to seek care in faith-inspired health facilities in Burkina Faso is the fact that the cost of care is lower than in public facilities (the other reason being that faith- inspired facilities appear to provide better quality of care). But how do faith-inspired facilities do it? While most faith-inspired facilities in Burkina Faso operate independently from the government, they are integrated within the national health system. This means that they receive various forms of support, in material resources (such as vaccines and mosquito nets) as well as human resources (e.g. training of personnel and salaries for a limited number of government health workers). They are also subject to quarterly inspections from district health officials. Nevertheless, faith- inspired facilities tend to receive only limited core support from the state to provide their services. The ability of the facilities to make quality care affordable for the poor must thus be maintained thanks to support in kind and in cash from other sources, such as religious groups and donors. Apart from documenting why faith-inspired facilities tend to be the preferred choice of provider among households living in areas where the facilities are located, the second objective of the paper is to discuss how the facilities, while receiving limited support from the state, do manage to make care more affordable. The paper is structured as follows. In order to provide context, section two summarizes findings from an analysis (including both statistics and regression analysis) of a nationally representative 2007 survey showing that cost is indeed a major obstacle to care, especially for the poor. Section three presents results from in-depth qualitative research conducted between April and July 2010 in six (two Protestant, two Islamic, and two Catholic) health facilities showing that households indeed value these facilities especially because of their ability to provide affordable care. Section four then provides administrative data from the facilities in terms of costs, budgets, and population served to show that support from religious groups as well as donors is key for the ability of the facilities to provide affordable care. Section five discusses some aspects of the collaboration between the facilities and the state. A conclusion follows. 103 QUANTITATIVE DATA AND ANALYSIS To what extent is cost a major factor reducing the demand for care in Burkina Faso? To answer this question, it is best to start with data from the nationally representative 2007 extended QUIBB (in French: ‘Questionnaire Unifié des Indicateurs de Base du Bien- être’) survey. This is the main multi-purpose household used in Burkina Faso for analyzing household well-being. While the survey unfortunately does not identify separately faith-inspired health care providers (nor do the Demographic and Health Surveys), it still provides useful information in its health module. Here, we are interested in cost as a barrier to the demand for care in the population. Table 1 provides the share of the population who sought treatment for illnesses or injuries. Among those who were ill or injured in the past 15 days, 62.4 percent consulted with a health service provider of some type (broadly defined, including traditional healers), but the proportion was only 53.1 percent in the lowest quintile of well-being (as measured through an index of wealth based on the assets owned by households), 70.7 percent in the highest quintile. A large majority of those in the wealthiest group (62.5 percent) did not seek health services because they chose instead to self-medicate, and that was also the main reason for the poor not to get a consultation, although for a lower share of the population in that group (53.8 percent). But the second most important reason for not seeking care was cost, for more than a third of the population overall, but less so in the top quintile. Other reasons for not seeking care included the fact that it was perceived as not necessary (16.5 percent of those not seeking care), or the fact that the facility was too far away (8.0 percent of those not seeking care). As suggested by these data, cost is a major obstacle to care, especially for the poorest groups. Table 1: Decision to seek care and reasons for not using health services, 2007 (%) Sex Residence Well-being - Quintiles All Male Female Urban Rural Q1 Q2 Q3 Q4 Q5 Share seeking care All 5.9 6.6 9.3 5.6 4.4 5.1 6.1 6.6 9.1 6.2 Population ill or sick 62.9 62.0 65.4 61.4 53.1 56.6 61.2 65.9 70.7 62.4 Reasons for no care Not necessary 17.9 15.8 16.0 17.0 18.8 16.1 12.2 21.0 16.0 16.8 Self-medication 56.4 54.7 61.7 53.7 53.8 56.4 54.7 49.9 62.5 55.5 Too expensive 32.0 34.9 29.0 34.9 35.4 37.3 38.3 34.8 21.7 33.5 Too far 9.4 6.9 0.5 10.2 14.1 9.0 5.4 10.0 1.2 8.0 Other reason 0.7 2.5 0.8 1.9 2.1 0.6 1.3 0.5 3.9 1.7 Source: Authors’ estimates using 2007 QUIBB survey. Another way to look at the data is to conduct regression analysis, and see which characteristics of households and individuals are correlated with the decision to seek care. This is done in table 2 through standard probit regressions, and we are reporting marginal effects (as opposed to coefficient estimates) for ease of interpretation. Given our focus in this paper on care provided in health facilities, we consider the demand for ‘formal’ health care when sick or ill, which thus excludes traditional healers. The analysis is done for the full sample at the national level, as well as for urban and rural areas separately, and for men and women also separately. 104 The first variable in the regression is the primary sampling unit leave-out-mean consultation rate when ill or sick. This is the share of those in a specific geographic area that are consulting when ill or sick, with the share computed on all individuals living in that small geographic area except the individual himself/herself. This variable captures a wide range of factors which are not observed in the survey, such as cultural attitudes towards facilities-based care, as well as the quality level of the care that is provided, since these influence the extent to which individuals in an area will seek care. As expected, the impact of that variable is positive, large, and statistically significant, with a value of one (indicating all other patients in an area apart from the individual seeking care when ill or sick) increasing the probability of seeking care for the individual by close to 40 percent in most cases. Next comes the quintiles of well-being of the households to which individuals belong. As was already apparent in the basic statistics presented in table 1, those in the wealthiest quintiles have a higher probability of seeking care, although the differences between the first four quintiles are in most case not statistically significant, perhaps because with about half of its population in poverty, and a substantial additional segment of the population at risk of falling into poverty, most of the individuals in the bottom three or four quintiles tend to have limited means to pay for care, and not only the poorest of the poor. There are also substantial differences in the likelihood of seeking care according to location in terms of regions, although not according to whether the household lives in rural or urban areas once one controls for regional location as well as for the distance to facility. The impact of the distance to the nearest health facility becomes statistically significant especially when the nearest facility is located at more than an hour away from where the household resides, in which case the likelihood of seeking care is reduced by approximately 15 percentage points in most cases (For research on geographical accessibility to care and its relationship to child mortality in Burkina Faso, see Schoeps 2010). As for the characteristics of the household head, and those of the individual who has fallen ill or sick, two main results emerge. First, there is some evidence that the likelihood of seeking care is lower when the household head is female, which may suggest a higher level of vulnerability for these households which limits their ability to pay for care (our quintiles of wealth are an imperfect measure to assess the vulnerability of many households since this is a stock variable, while shocks such as loss in incomes or other revenues may also affect the demand for care, and may be more likely to affect especially female headed households in a country such as Burkina Faso given that the absence of a male head suggests fewer potential earners in the household). In addition, the younger the child, the less likely that a visit to the health centers or health professional will be sought, possibly because small children fall sick more easily, with many episodes of such illness likely to be benign and not necessarily requiring a visit to a professional. 105 Table 2: Correlates of decision to seek formal care in past 15 days when ill or sick National Urban Rural Male Female dF/dx dF/dx dF/dx dF/dx dF/dx PSU LoM consultation rate 0.3963*** 0.1933 0.3965*** 0.4500*** 0.3802*** Quintile of well-being Poorest (Q1) Ref. Ref. Ref. Ref. Ref. Q2 -0.0699 0.0592 -0.0780 -0.1007 -0.0256 Q3 0.0040 - 0.0270 0.0081 -0.0156 Q4 0.0765 0.1393** 0.0734 0.0620 0.0747 Richest (Q5) 0.1019* 0.1997*** 0.0804 0.0187 0.1744** Region and location HB Ref. Ref. Ref. Ref. Ref. BM 0.0923 0.0298 0.1955*** 0.0654 0.1164 SH 0.1495** 0.0984 0.2513*** 0.0931 0.2031** East 0.0135 -0.1107 0.1370* -0.0373 0.0858 SO 0.1339** - 0.2398*** 0.0719 0.2166*** CN 0.1545** 0.1867** 0.2477*** 0.1909** 0.0885 CO 0.1755*** 0.0220 0.3039*** 0.1141 0.2485*** PCL 0.0411 - 0.1669** 0.0077 0.0885 Nord 0.1505** 0.0073 0.2590*** 0.0835 0.2047** CE 0.2143*** 0.0134 0.3557*** 0.2054*** 0.2395*** Centre 0.1001* -0.0470 0.3563*** 0.0732 0.1590** Cas 0.0128 0.0602 0.0657 -0.1630 0.1735 CS 0.1490** - 0.2578*** 0.1050 0.2188** Urban location 0.0632 - - 0.0950 0.0579 Characteristics of head Age -0.0030 0.0133 -0.0098 -0.0125* 0.0075 Age squared 0.0000 -0.0001* 0.0001 0.0001 -0.0001 Female -0.0916* 0.0608 -0.1858*** -0.1834** -0.0393 No education Ref. Ref. Ref. Ref. Ref. Primary education 0.0189 -0.0453 0.0707 0.0502 0.0125 Secondary education 0.0605 0.0345 0.0549 0.0315 0.0741 Higher education -0.0150 -0.0094 - 0.1127 -0.1230 Characteristics of individual Age -0.0692*** -0.0762*** -0.0660*** -0.0718*** -0.0671*** Age squared 0.0042*** 0.0045*** 0.0040*** 0.0040*** 0.0044*** Female -0.0274 -0.0340 -0.0105 - - Handicapped -0.0712 -0.3041 0.0382 -0.4291*** 0.1848 Below 16 and orphan -0.0916 -0.0374 -0.1341 -0.0557 -0.1626* Time to nearest facility 0-14 minutes Ref. Ref. Ref. Ref. Ref. 15-29 minutes -0.0964** -0.1004* -0.1027* -0.0631 -0.1118* 30-44 minutes -0.0355 -0.0013 -0.0506 -0.0046 -0.0324 45-59 minutes 0.0015 -0.0926 0.0200 -0.0808 0.1049 60 minutes or more -0.1582*** -0.1788 -0.1529*** -0.1687*** -0.1181* Number of observations 1505 467 1034 784 721 Source: Authors’ estimation. Probit estimation reporting marginal effects. Note: Levels of statistical significance: *** p<0.01, ** p<0.05, * p<0.1. What can be concluded from this brief analysis of the survey data? Perhaps the most important conclusions from a health policy point of view are the fact that cost remains a barrier for many to seek care, while lack of access in terms of distance to facilities is also a constraint. Lack of access however seems to affect only a minority of the population, 106 that is, those located more than an hour away from the facility. By contrast the issue of cost is a more widespread problem if one compares the lower demand for care in the bottom four quintiles or eighty percent of the population to the higher demand observed in the top and wealthiest quintile. This conclusion is also supported by the basic statistics presented earlier in table 1, where cost was mentioned as the main reason for not seeking care by 33.5 percent of the population nationally, as compared to 8.0 percent mentioning the distance to facilities as the reason for not seeking care. QUALITATIVE DATA AND ANALYSIS In order to dig deeper into the issue of cost as a constraint to care, and of the role of faith- inspired facilities in making care affordable for the poor, we conducted in-depth qualitative research between April and July 2010 in six (two Protestant, two Islamic, and two Catholic) health facilities. Three of the facilities were very small with fewer than 12 workers, two were mid-sized with 30-40 workers, and one was a larger hospital with 213 workers. The three rural facilities are designated by the state as a CSPS (clinic) and the three urban facilities are considered as CMAs (hospitals). The facilities were selected by health ministry officials and the research team on the basis of their being located in areas where both public and faith-inspired healthcare options are available, so that we could compare both types of facilities. This would allow respondents to discuss the advantages and disadvantages of the different facility types. A semi-structured questionnaire was used to interview 48 patients in the six facilities (eight in each facility). This is admittedly a small sample size, but we were more interested in-depth analysis than statistical representativeness. Because assessments of facilities by respondents were converging to a very large extent, we are confident in the findings obtained from the interviews. The questions focused on patients’ views and motivations concerning healthcare, their evaluation of the faith-inspired health centers and the comparison with their experiences in public centers, and the way in which cultural and religious values shape decision-making. Opportunistic sampling was used and the sample was gender stratified with an equal number of male and female patients. An attempt was also made to stratify the sample by religion, according to the percentages of different religious groups attending each facility (if half of the patients were Muslim then 4 respondents were Muslim, if three-quarters Muslim, then 6 respondents were Muslim). Interviews were conducted by a Burkina Faso based research team in French and local languages (Moore, Dioula, Peulh), depending on the primary language of the respondent. Semi-structured interviews were also conducted with two health center leaders including the director and a doctor or head nurse. A total of 24 focus groups in faith communities in one rural and one urban area provided an opportunity to further investigate the intersections between faith and health. Focus groups were conducted in Ziniare, a rural town 40 kilometers from Ouagadougou, and in the city of Bobo Dioulasso. The groups were conducted with Muslims, Catholics, Protestants and Traditional Religious groups, and within each religious community, men, women and religious leaders each had their own discussion. 107 The qualitative data confirm that cost is a major obstacle to care, but in addition, it suggests that faith-inspired facilities have a comparative advantage in this area versus public facilities through their ability to provide lower cost as well as higher quality services. The question of the quality of the services is discussed in a companion paper (Gemignani and Wodon 2011). Here, we focus on cost, documenting both how high costs reduce the demand for care, and how faith-inspired facilities appear able to reduce at least some of the cost burden for their clientele. The majority of respondents in our fieldwork explain that they attend the faith-inspired facility in large part because these providers offer care at a lower cost than secular clinics and hospitals. They describe significant cost differentials for both services and medication. When asked about the main advantage of faith-inspired health providers as compared to public facilities, 54 percent of respondents mentioned lower cost of care. This response was common across gender, religion and location (rural/urban). It also came up strongly in testimonies provided by respondents: “The sisters do not threaten the patient. Instead they help the patient because they do their work for God. As soon as we arrive, very sick, they start to look after us. There are four of them and all four put themselves at the service of the patient. They are not quiet, as long as the suffering of the patient continues. Then, two or three days later when your health has improved, that is when they tell the family the cost of the treatment. Elsewhere you cannot have emergency care without paying. It is sure that you will die then, because no one pushes themselves to take care of you and it seems that death no longer means anything to them� (female patient, Catholic clinic). Quality was also mentioned as very important and a comparative advantage for faith-inspired facilities. But in addition, another perceived advantage of faith-inspired centers is that they provide counseling or other services for those who would like to benefit from them: “For one year since I’ve come to this center for maternity issues, I haven’t spent anything. My child and I are nourished and cared for free of charge. I had 11 children and 9 have died. Currently I receive care for one of my surviving children. The two of us are taken care of at the center. The pastor meets with me regularly for prayers and provides counsel for my maternity problems. I didn’t have any of these advantages at the other CSPS’s which weren’t able to do anything to save my children…. It is ignorance and the lack of visits to good health centers which cause us to have these problems� (patient at Protestant clinic). Some respondents felt a lack of transparency among some public providers and suggested that excessive prices are charged for certain supplies and services. Others list the many costs incurred in attending a health center, and the lack of payment options for the poor. For some, the public health system seems to represent further health and financial woes rather than a place which offers viable treatment options. The following illustrate respondents’ views regarding healthcare costs: “When the women go to give birth in the CSPS, the government says that it is free but… one must pay for gloves, bandages, compresses, etc. Here, even if one pays, it is not in an exaggerated way because the price is low and within everyone’s reach� (male patient, Muslim clinic); “Last year I sent my wife to give birth at [CMA]. They made me spend a lot of money. They said that it is a public hospital directed by the state and that the ministry covered most of the expenses, 108 but that was not the case because I had to purchase many products. And now I prefer to come here because not only is it closer to my residence but also when I come here, the care and the drugs cost less… I no longer want to visit [CMA], because there one does not seek to know if you are poor or not, and one does nothing but prescribe you ordinances without knowing if you have the means. So if you go there and you have nothing, you are just going there to die� (male patient, Muslim clinic); “I once stayed at the public hospital… Every morning I paid the ordinances. There was no progress but I noted that the health personnel were swindling me. Every day they required me to purchase products sold at a high price. That really marked me, and I haven’t returned to that hospital for care. Here, the health personnel are not corrupt, and they are friendly and respectful� (female patient, Protestant clinic); “My husband no longer complains of the medical ordinances because, for 6 years now, I have seldom received them. He believes that the CSPS in our area exists simply to sell its drugs, that it is not a medical centre but a business established in order to market drugs� (female patient, Protestant clinic). There is thus a perception – which may or may not be valid - that some public facilities are affected by petty corruption, whereby the prices of consultations or drugs are inflated for the benefit of staff. It is also perceived that public facilities simply function as sellers or services in a market, without as much attention paid to patients and especially the poor. Whether this is true or not is unclear, and it may be that the association of faith- inspired facilities with religious aims gives them an advantage in terms of not being likely to be perceived as taking advantage of patients. But the difference in perceptions is still clearly there: “Unlike the public clinics, one does not prioritize money in the religious clinics, because it is God who is central to the healing. Thus the services are given in a way that is social and humane. It is to help the poor, the disadvantaged groups� (female patient, Protestant clinic); “If you go to the CMA, people might as well start to cry for your corpse. They do not have pity even for people who are seriously ill. First they want to know how much you can give to the nurse for dealing with you. It is like at the market, between the salesperson and the customer… Sometimes the disease is very serious and [the sisters] advise the family to go to the CMA and we are sad because we know what awaits us. For this reason we pray to God that the sisters will be able to care for the patient so that all is limited to the center� (female patient, Catholic clinic); “Over there [CMA], the health personnel are dishonest. They steal from the patients by charging high prices for consultations and medicines. This practice is common at [CMA]� (male patient, Muslim clinic). There are also indications that while all services must be paid for in public facilities, this is not always the case in faith-inspired facilities, where efforts are made to keep care affordable for the poor, which means sometimes giving them a break on payment for care. This also means that the same service is provided to the poor and the better off in faith-inspired facilities, which may not always be the case in public facilities. A few examples illustrate the perceived differences well: “This pharmacy has become an important resource for everyone. One gets their ordinances at [CMA], but gets the drugs in the sisters’ clinic. Otherwise, certain ordinances are never going to be paid� (male patient, Catholic clinic); “I am not the only one who complains about other CSPSs 109 because of their attitude toward patients and the negligence in consultation and treatment…. In some CSPSs, we also find that one is treated differently, depending on whether one is wealthy or poor� (male patient, Catholic clinic); “In a religious medical center, they work for God and those who work for God know that they must help the poor…The places where God’s name is absent, you go only because you don’t have the choice and you know what awaits you. It is necessary to pay for everything; it is necessary to wait a long time; it is necessary to bribe the personnel if you want them to receive you or propose quality care to you. With us peasants, they often like to tell us to go sell our livestock and bring the money and we will look after you. And if we don’t agree to do this, we will die� (female patient, Catholic clinic). The issue of cost has important gender dimensions, since many women are not able to afford healthcare on their own and depend on the willingness of their husbands to pay for their health needs. For example, one woman who had travelled an hour to the faith- inspired clinic described the inability of many women in her village to afford the services at the local CSPS as follows: “At our health center, we find that behind each instance of health care is hiding a financial expenditure so that our husbands prevent us from attending the centers. It is only when the situation worsens that they take the woman there. The journey is difficult. One must go by bicycle, [donkey] cart, or motorbike and this is why we lose our pregnancies� (female patient, Protestant clinic). Another woman explained: “Before, women didn’t give birth in the centers due to the costs. Then, the sisters met with groups of women and told us that they will no long ask expectant mothers to pay for [prenatal] consultations. Only at the time of childbirth is it necessary to pay 900F and for one year after childbirth, the mother and the newborn will receive free care. When we had this information, the problem of childbirth in our village was solved. Now all is done in the center� (Catholic clinic). The lower costs of care thus improve husbands’ support for the care of their wives. Indeed, several women described how their husbands who once complained of their medical expenses now approved them and even encouraged clinic attendance: “The men have noticed that we have fewer health problems now that we are attending the health center during our pregnancy and the first weeks after the birth of the child. Our husbands accompany us and some witness our consultation and pay attention to the message given by the health worker� (patient at Protestant clinic); “When I used to attend [CMA], my husband did not accompany me. In contrast, he himself advises me to visit the sisters’ health center and very often accompanies me there� (patient at Catholic clinic). Many households do not go to the health centers due to both cost and cultural practices (such as relying on traditional healers, or simply not seeking care until absolutely necessary). By reducing the cost barrier, faith-inspired facilities also help reduce cultural barriers to the demand for care: “There are cultural practices that have a negative effect on the health of the population. For example, there are many people who do not see the doctor until they are gravely ill….Here we have the habit of saying ‘When it gets hot� which means that when the person is in risk of death, that is, when we take them to the hospital. For some people, visiting the hospital is the last resort. This is because of certain beliefs, sociocultural biases, and also because of the lack of financial means to buy medicines at the pharmacies� (female patient). And again, as already mentioned, the 110 affordability of care also helps in changing men’s attitudes towards care: “When it was necessary to go to the CMA of [town], I can say that very few among us went there for the consultations because our husbands would say that we want to bring him problems. As long as a woman is not confined to bed, she is not regarded as a patient. But with the opening of the center of the sisters, everyone knows that it is free. It is known that there is a sister who will take care of us and our husbands are not opposed. It is when money is needed for care that the battles erupt in the family. This is why a suffering woman, even if she is pregnant, is afraid to alert her husband. He will only see the financial expenditure� (Catholic clinic). ADMINISTRATIVE DATA AND ANALYSIS Budget analysis The qualitative analysis presented in the previous section makes it clear that faith- inspired facilities are able to provide services to the poor at lower cost than public facilities. At the same time, faith-inspired facilities tend to receive less support from the state for their operating and other costs than public facilities. This begs the question as to how faith-inspired facilities can remain financially sustainable with their practice of subsidizing the poor. In this section, we use partial data on costs and revenues for the faith-inspired facilities visited during the fieldwork to try to better understand how they are indeed able to implement what appears to be a preferential option for the poor (this is a Christian term, but focusing on the poor is also part of Islam). A detailed comparison of the costs of faith based and public health care was not included as part of this study. But we did ask administrators in faith-inspired centers to provide the cost of a general consultation. These ranged from 150–1000 F CFA. This is lower than the 2000 F CFA often charged at the public hospitals, but the same or slightly more than the fee at the public CSPS, which is approximately 200 F CFA (the public fee for simple delivery, 900 F CFA, and complex delivery, at 1800 F CFA, are also similar to the fees in faith-inspired centers). On the other hand the faith-inspired centers we visited did appear to provide free care to those in need – often waiving consultation fees. It was also apparent from our discussions that the basic fees for consultation and deliveries are not the only cost of concern. Respondents emphasized a multitude of additional charges due to the need to purchase medicines and supplies, for example for childbirth. Basic supplies and generic medicines at the faith-inspired centers were apparently available at low prices and again, the poorest were not required to pay. Respondents also pointed out that since medical personnel are focused on the patient’s ability to pay, they avoid writing long ordinances listing numerous medications that are not essential to the treatment. It is also useful to consider some of the operating costs (e.g. employee salaries) at the faith-inspired centers, in order to determine if providing higher quality care entails larger costs which then must be passed on to the clients. Table 3 shows salaries for various categories of faith-inspired and public health workers. It is important to note that the salaries for public sector employees are starting salaries, while the faith-inspired salaries are based on the average of current salaries earned in the various categories. The average length of employment for the faith-inspired health center employees is 9 years. Given that 111 inflation is limited in Burkina Faso due to the peg of the F CFA with the Euro, even if raises are provided with seniority, `we would not expect salaries for government health workers to increase extremely rapidly over time. Table 3: Health worker monthly salaries in faith-inspired and public facilities Average Average Average Starting Salary: Salary: Salary: Salary: All faith-inspired Protestant Muslim & Catholic Government Head of clinic 383,362 436,586 356,750 -- Doctor 284,300 350,000 262,400 146,000 Health officer 169,382 169,382 -- 140,000 Registered midwife 122,691 138,885 98,400 110,000 Registered nurse 114,094 146,111 88,480 110,000 Licensed nurse 94,912 122,824 67,000 100,000 Itinerant health agent 92,088 134,175 50,000 70,000 Nurse’s aide 98,238 98,238 -- -- Orderly 53,268 66,428 31,333 55,000 X-ray technician 160,000 160,000 -- 110,000 Laboratory technician 113,186 150,780 -- 110,000 Midwife (matron) 100,994 100,994 -- -- Birth attendant 72,500 -- 72,500 70,000 Caregiver (infant) 30,004 30,017 30,000 -- Accountant 101,931 138,261 65,600 -- Other (cashier, guard) 69,712 81,625 47,375 -- Source: Compiled by authors. Note: Starting salaries are provided for government workers, while workers in faith-inspired centers have an average of 9 years experience. The starting salary for government doctors represents the average of “specialist� (160,000 F CFA) and “generalist� (132,000 F CFA). The list of salaries from the faith-inspired centers includes 94 workers, but the total number of workers in these six centers is 204 (due to the lack of easily retrievable records, the 6 centers did not provide a complete salary list). Only the workers whose salaries are paid directly by the faith-inspired centers are included in these figures and in the statistics above (not those paid by the state). As shown in table 3, salaries for doctors in the faith-inspired centers after an average of nine years of experience were significantly higher than starting salaries in public health centers. There were no doctors in the Catholic centers or the Protestant CSPS, but the doctors in the Muslim centers earned 80 percent more than starting government salaries. In the Protestant hospital, doctors earned more than double the government salary (when we inquired about these large differences, we were told that this was due to the long hours the private physicians were required to work in the hospital). However, for other categories of health workers, the divergence from government salaries was much smaller. In the Muslim and Catholic centers, the salaries of workers such as midwives and nurses appear to be less than the official government starting salaries. For the Protestant health centers, there is a modest increase in salaries. For example, registered nurses and midwives (Infirmier d’Etat and Sage Femme d’Etat) made 33 percent and 26 percent more than government starting salaries. Overall, it is this not clear that faith-inspired facilities would enjoy a comparative advantage (or disadvantage) in terms of cost structures due to salary structures that would differ markedly from those of government workers (although the fact that some nurses do not benefit from the same salaries does matter, as discussed below). 112 Tables 4 and 5 provide some idea about broader costs of operations in faith-inspired facilities by presenting estimates of annual revenues in relation to capacity for five of the six facilities. The smaller clinics earning only about 9 million F CFA annually (approximately $20,000) have 10-12 staff, 20-27 beds, and are able to serve about 150 people per week or 8,000 clients annually, which is very substantial and highly cost effective. This ability to reach so many with very limited financial resources is largely due to the sisters’ financial support as well as the donations of equipment and medicine from congregations or other support groups. The mid-size facilities have revenues around 120 million F CFA (around $270,000) and have 32-42 staff members, 40-63 beds and serve about 800 clients per week or 42,000 per year. Finally, the larger hospital has about F CFA 900 million in total revenues (approximately $1,800,000), 50 beds, and 160 permanent staff and serves some 2102 clients per week or 109,304 annually. This larger hospital thus has two to three times more patients than mid-size facilities but seven to eight times more revenues. The relatively high revenues are possibly related to pharmacy, laboratory and imaging services. Since the hospital could not provide a record of total patients served across the various departments, an estimate was calculated from the annual number of patient consultations and medical procedures such as childbirth and surgeries. Those who are filling a prescription, or referred to the hospital for an x-ray or laboratory test are not included in the number of patients served. In addition, the cost per patient may be higher in a larger hospital due to more complex procedures It is also important to note that in recent years a significant portion of revenues for this specific hospital has gone to construction projects and the expansion of the health center facilities. In order to check if these estimates provided to us by the centers made sense, we computed the ‘cash cost’ per patient in the five centers by simply dividing estimates of total revenues (not including salaries paid by the government or living stipends paid by congregations) by the number of patients served each year. The two clinics with the lowest cash costs were the Catholic facilities run by sisters, at respectively F CFA 1304 and F CFA 1803 per patient (or approximately three to four US dollars per visit). The fact that these facilities had the lowest cost per patient is not surprising given the availability of sisters not paid in the same way as other health professionals. Two other facilities had cash costs near F CFA 3000 per patient. Only the last facility had a higher cost per patient (F CFA 8180), but as mentioned above, this is likely due to the more complex procedures offered in the hospital as well as the revenues from the pharmacy and laboratory, and the fact that part of the resources have been used to expand facilities. The fact that the cash costs per patient for the first four facilities are of a similar order of magnitude is reassuring for the validity of the data. 113 Table 4: Health center data on revenues, staff size, and number of patients served Faith Amount Support Support Total No. No. No. No. Total No. of No. of No. of No. “Cash received from relig. from revenues MDs nurses other state Staff patients patients patients of cost� from patients groups and state and workers workers per per per beds per (F CFA) other orgs. health week week week patient (F CFA) officers (Male) (Fem.) (total) (FCFA) Musl. 54 million 75,400,000 4 nurse 129,400,000 4 9 15 5 33 286 633 919 40 2708 salaries + 4 nurses Cath. 9 million Sisters -- 9,000,000 + 0 3 9 0 12 42 54 96 27 1803 receive sisters’ support, financial room & support board, etc. Cath. 8 million Sisters 1 nurse 8,000,000 + 0 3 6 1 10 42 76 118 20 1304 receive salary 1 nurse + support. sisters’ support Prot. 107 million 5,141,462 600,000 112,739,860 0 15 27 0 42 249 442 691 63 3136 Prot. 800 million 94,086,027 50 894,086,022 7 68 38 50 163 + -- -- 2102 50 8180 worker + 50 state 50 salaries workers temp. workers Source: Compiled by authors. 114 It is striking to see in table 4 how large support from religious groups and other organizations is in comparison to cost recovery fees from patients and support from the government. Consider for example the first (Muslim) clinic. Its support from religious and other groups is equivalent to 1.4 times its revenues from patients, while support from the state is rather limited given the salaries reported in table 3. Support from the state is also very limited in the next three facilities in the table, and is substantial only for the last facilities through payment of salaries for 50 health workers. In the Catholic facilities, as already mentioned, the fact that the sisters tend not to be paid as professional and received support from their congregations helps a lot as well. Overall, support from religious groups as well as others donors helps in enabling the facilities to provide subsidized care for the poor, given that such support is large versus payments by patients. Table 5 compares the overall budget of the facilities with their costs. Note that for some facilities, revenues are being used to expand facilities. All facilities tend to benefit from donated medicine, equipment, and supplies, although we do not have data on the value of such transfers. It is also interesting that several facilities seem to have a budget for social assistance. Possibly those funds are used to make care more affordable for specific groups, but it is also likely that some of the funds are being used to make additional services available to some families (such as food prepared for young children when parents consult), as suggested by the qualitative work. Table 5: Overview of facility budgets Facility 1 Facility 2 Facility 3 Facility 4 Facility 5 Resources (annual): Monetary 129,400,000 9,000,000 8,000,000 112,480,252 807,872,000 +4 nurse + Financial + 1 nurse + 50 state salaries support for salary worker salaries sisters Non-monetary Donated Donated Donated Donated Donated medicine, medicine, medicine, medicine, medicine, equipment & equipment & equipment & equipment & equipment & supplies supplies supplies supplies supplies Expenses (annual): Salaries 14,400,000 6,612,000 5,400,000 56,796,403 192,000,000 Supplies 500,000 300,000 50,000 2,621,208 243,000,000 Utilities 600,000 1,000,000 200,000 1,173,425 20,000,000 Maintenance 300,000 100,000 75,000 1,890,119 10,000,000 Other 37,785,600 150,000 (social 400,000 49,806,695 118,000,000 (misc. assistance) (social (construction) including assistance) 34,000,000 social (social assistance and assistance) new 4,000,000 construction ) (training) Source: Compiled by authors. 115 Collaboration with the state During an inauguration at a Protestant medical center in 2010, then-Prime Minister Tertius Zongo stated: “What is most important is to see that the private or religious [health centers] can provide quality healthcare alongside the efforts of the state� (Burkina Faso Prime Minister’s Office 2010). This is indeed the spirit in which faith- inspired facilities seem to be working. The analysis of our data suggests that faith- inspired facilities perform a valuable service for the communities in which they are operating. According to their leaders, the centers were actually established in areas that had (at the time of creation) limited access to the health system. There is also a strong focus on providing services to the poorest even when facilities are located in urban areas, and this is found across the different types of providers (Catholic, Muslim and Protestant). One clinic leader described their target zone as follows: “This health center is located in the densely populated neighborhoods. We also serve the peripheral or unincorporated area which is densely populated as well. In the entire zone, there is only one public CSPS. This explains the large numbers who visit this clinic. Also the majority of the people who live in the areas that I have just described are poor and live in conditions that are not at all decent. In the unincorporated areas, there is no infrastructure including tap water, gutters and garbage disposal… The patients’ economic situation is catastrophic because the families depend on odd jobs in the informal sector - selling various items, masonry, itinerant salesman of cigarettes - and on small agriculture… Most are without school instruction. They are unaware of all the rules of hygiene, the questions of family planning, and the advantages of [Western] medicine.� All of the health center leaders provided similar descriptions of their target population. Several described a greater focus on women and children, with emphasis put on maternal care and preventative care for infants and children. Most administrators also said that they are integrated into the national health system and they tend to be fairly satisfied with their relationship with the state. One official at a public clinic, however, described realistically a sense of competition between public and faith-inspired services, and a negative view toward the free or low cost care that faith-inspired centers are able to offer to patients and which may in some cases significantly lower the client base in the public sector. That some competition exists is indeed clear. Still, while faith-inspired facilities benefit from support from religious groups and other donors, the health center leaders strongly emphasize the importance of support provided by the state. According to the administrators, the provision of personnel is the most important support they receive from the state. The numbers of state personnel varied according to the size of the facility, with one staff member provided to a small clinic, five state employees at mid-sized clinics, and fifty state employees at the large faith-inspired hospital. All of the leaders interviewed stated a desire for more state support in the area of personnel. For example, a leader at a Muslim clinic stated “More midwives and nurses trained by the state should be provided to us. Unfortunately this is not done and the situation creates blockages in the operation of our services. It also creates an overload of work and our personnel is always overwhelmed by the large numbers of patients who do not always understand and tolerate the long lines and waiting periods.� 116 Leaders also described their desire to improve and expand their facilities and the ongoing need for equipment and supplies in many areas of care. At all the smaller clinics, they described “making do� with what is on hand. One Protestant clinic was well funded, and a Catholic clinic had recently secured financing for a significant expansion, but for the most part, there is a notable lack of resources. Some clinics reported lacking even basic supplies such as thermometers and blood pressure monitors. One clinic described their futile attempts to attain a vaccine refrigerator. A Catholic clinic known for its pre- and postnatal care in the villages described the lack of basic equipment in the maternity ward. Nearly all of the leaders to whom we spoke reported positive relationships with the government: “In recent years, the government has made an effort to work with us and our collaboration is fruitful. They recognize our strengths and although they don’t provide many resources for our functioning, they do a lot.� Although they were not provided with large amounts of human and material resources, what they did receive was well appreciated and they had very few negative comments about the collaboration itself. In addition to providing the health centers with state health workers, the state provides other resources to certain facilities including medical supplies (e.g. vaccines, mosquito nets), equipment (e.g. refrigeration systems), support for the nutritional health centers (CREN), and periodic trainings for personnel. Trainings were viewed as very useful for building the skills and knowledge of the staff: “I really appreciate the successful collaboration between the district and the private medical structures. There is a desire to work together to answer the health problems of the population. The [Medical District Head Physician] is very attentive to the quality of services of this center and said that he wants to provide us with qualified health personnel.� Government inspection teams regularly visit the health centers and all of the leaders said that they viewed this in a positive way: “Each quarter we receive an inspection team from our health district. They come in order to supervise us and provide advice on operations. They discuss our strong points and our weak points, and then we try to apply their recommendations. This works out well and I appreciate this type of support…. We have discussed many of our clinic’s problems with them. Their concern to communicate with us is already a type of collaboration and it is positive. The government has set themselves up like a partner who does the work of evaluating our activities. This enables us to correct our weaknesses and improve our services.� In return for support received, the health centers provide monthly, quarterly, and annual reports of their operations. The centers are seen as providing valued services that reach the poor and this aspect of their work is recognized by state health authorities. When asked about how they would improve their relationship with the state, about half of the leaders said that they would like to see greater support from the state in the future. More frequent training was one suggestion and many of the leaders hoped that more personnel could be provided: “The only negative aspect of our collaboration is the level of support. We are a social center with a non-lucrative goal. Our goal is to build and equip the medical center. But we must pay most of the personnel... With our thin resources we pay the water and electricity bills and most of the workers... I have 33 workers here and 5 are 117 paid by the state… We are the only nongovernmental structure to provide care under SONU (free childbirth services)… Other nongovernmental structures refuse to assume responsibility for this care. We try to satisfy the needs of the population of the poor and should receive more support through the supply of personnel.� One larger clinic also felt that it is unable to realize its full potential for health care provision because state policy does not allow a CSPS to provide certain kinds of services designated for the district hospitals and surgery centers. Examples included blood transfusion and cataract surgeries. Their future plans include training some of their personnel to become doctors so that they can change their status (from CSPS to CMA) and expand the provision of services. CONCLUSION This paper has described the role of faith-inspired health providers in Burkina Faso. Quantitative results from the national survey were combined with in-depth qualitative research and administrative data in order to understand the successes of faith-inspired services especially in terms of their ability to subsidize care for the poor, as well as some of the current challenges they are facing. Some of the topics covered here include utilization of formal health care and barriers to access, the unique contributions of faith- inspired services, the level and nature of collaboration with the government, and some of the external and internal limitations to healthcare provision. Faith-inspired healthcare in Burkina Faso is still in an early phase of development, but the contributions of the religious organizations are noticed not only by the state but by the many patients who extol the benefits of their approach. One unique aspect of this in Burkina Faso is the fact that a majority of faith-inspired health centers appear to be reaching poor members of society and that they are viewed as equally or more affordable than even the most basic public health centers (CSPS), which itself is made possible through the support that the centers get from religious groups and other donors. This does lead to some competition with public facilities, but it also increases the availability and quality of the care being provided. The qualitative research suggests that patients place a great deal of emphasis on the quality of care they receive in the health centers. Interpersonal relations in the faith-inspired centers are viewed as very important and many contrast the sense of compassion, trust and “bonds of friendship� with the more impersonal and sometimes hostile environment of public clinics and hospitals, as patients perceive it. While the administrators of faith-inspired health centers view their basic support from the government in a favorable light, they also state their need for further assistance in order to meet their expenditures for staff, training, medical supplies and other costs. They view their role in addressing the needs of the most disadvantaged members of society as a calling and service to others and in so doing they operate with a great deal of flexibility. At the same time, many note the disparities between available resources and demand for services and they also hope to expand their ability to meet the many health needs in the communities. As one director stated, “the resolution of one need creates another need.� Acknowledging the important contributions of these institutions and increasing their 118 funding and other supports through enabling policies in a way that would not be seen as threatening for public facilities could make a significant impact toward improved community health in Burkina Faso. REFERENCES Baltussen, R., Y. Yé, S. Haddad, and R. Sauerborn. 2002. “Perceived Quality of Care of Primary Healthcare Services in Burkina Faso.� Health Policy and Planning 17 (1):42-48. Banda, M., E. Ombaka, S. Logez, and M. Everard. 2006. Multi-Country Study of Medicine Supply and Distribution Activities of Faith-Based Organizations in Sub- Saharan African Countries. Geneva: World Health Organization and Ecumenical Pharmaceutical Network. Bodart, C., G. Servais, Y. Mohamed and B. Schmidt-Ehry. 2001. “The Influence of Health Sector Reform and External Assistance in Burkina Faso.� Health Policy and Planning 16 (1):74-86. De Allegri, M., V. Ridde, V. Louis, M. Sarker, J. Tiendrebeogo, M. Yé, O. Muller, and A. Jahn. 2011. “Determinants of Utilisation of Maternal Care Services After the Reduction of User Fees: A Case Study from Rural Burkina Faso.� Health Policy and Planning 99 (3):210-18. Faye, S. L. 2008. “Becoming a Mother in Senegal: The Experience of Motherhood in a Setting of Social Injustice and Health Service Failures.� Cahiers D’études et de Recherches Francophones/Santé 18 (3):175-83. Foulon, G. and R. Some. 2005. “Quel Système de Financement de L’accès aux Soins des Populations Dans Les PED: Le cas des Districts de Santé au Burkina Faso.� Mondes en Développement 33:99-110. Haddad, S., A. Nougtara, and V. Ridde. 2004. “Les Inégalités D’accès Aux Services de Santé et Leurs Déterminants au Burkina Faso.� Santé, Société et Solidarité 2:199-210. Hampshire, K. 2002. “Networks of Nomads: Negotiating Access to Health Resources Among Pastoralist Women in Chad.� Social Science & Medicine 54(7):1025-1037. Marschall, P., and S. Flessa. 2009. “Assessing the Efficiency of Rural Health Centers in Burkina Faso: An Application of Data Envelopment Analysis.� Journal of Public Health 17:87-95. Nikiema, B., S. Haddad, and L. Potvin. 2008. “Women Bargaining to Seek Healthcare: Norms, Domestic Practices, and Implications in Rural Burkina Faso.� World Development 36(4):608-624. 119 Nitiema, A., V. Ridde, and J. Girard. 2003. “L’efficacité des Politiques Publiques de Santé Dans un Pays de L’Afrique de L’Ouest: Le cas de Burkina Faso.� International Political Science Review 24(2):237-256. Pokhrel, S., M. De Allegri, A. Gbangou, and R. Sauerborn. 2010. “Illness Reporting and Demand for Medical Care.� Social Science & Medicine 70:1693-1700. Ridde, V. 2007. Equité et Mise en Oeuvre des Politiques de Santé au Burkina Faso. Paris: Editions l’Harmattan. Ridde, V., M. Yaogo, Y. Kafando, O. Sanfo, N. Coulibaly, A. Nitiema, and A. Bicaba. 2009. “A Community-Based Targeting Approach to Exempt the Worst-off From User Fees in Burkina Faso.� Journal of Epidemiology and Community Health 64 (1):10-15. Schmid, B., E. Thomas, J. Olivier, and J. R Cochrane. 2008. The Contribution of Religious Entities to Health in sub-Saharan Africa, Cape Town: African Religious Health Assets Programme. Schoeps, A., S. Gabrysch, L. Niamba, A. Sié, and H. Becher. 2011. “The Effect of Distance to Health-Care Facilities on Childhood Mortality in Rural Burkina Faso.� American Journal of Epidemiology 173 (5):492-498. United States Agency for International Development (USAID). 2009. Country Health Statistical Report: Burkina Faso. Washington, D.C.. Wodon, Q. 2013. Faith and Human Development: Education and Health Services in Africa, New York: Palgrave MacMillan. World Bank. 2009. Country Assistance Strategy for Burkina Faso for the Period FY 10- 12, Washington, DC. 120 CHAPTER 8 BENEFIT INCIDENCE OF PUBLIC HEALTH SPENDING FOR PUBLIC AND FAITH-INSPIRED PROVIDERS IN GHANA Harold Coulombe and Quentin Wodon 22 The World Bank This paper uses the fifth round of the Ghana Living Standards Survey collected in 2005-2006 to conduct a benefit incidence analysis of public spending for health. District-level financial data on public transfers are combined with household survey data on the use of various types of facilities by the population to assess whether public health spending reaches equally various segments of the population. The estimates of benefit incidence are presented separately for public and faith-inspired facilities, given that the later also benefit from public funding. The analysis suggests substantial variation in transfers and unit costs by districts, with higher costs in areas with the lowest and highest poverty measures, and lower costs in-between. Public health funding is also found to be regressive, in large part because hospitals and clinics still benefited the better off more than the poor, whether they are operated by the government or faith-inspired providers. INTRODUCTION Historically, as is the case for many other sub-Saharan Africa countries, Ghana’s public health spending was concentrated in urban areas and tertiary-level health facilities. In recent decades however, significant reforms have been undertaken in order to improve access to health services among the poor, as well as make the services provided more affordable (e.g. Salisu and Prinz 2009). Organizational and fiscal decentralization reforms were implemented among others through the creation of Budget Management Centers designed to facilitate transfers of resources from the Ministry of Health to districts-level authorities. Higher emphasis was placed on provision of primary care through the development of Community Health and Planning Services facilities. And a fundamental reform of funding for health care was initiated with the creation of the National Health Insurance Scheme and its exemption policies designed to reduce barriers to access for the poor. Today health remains a major priority for the Government of Ghana, which has positioned Human Development as one of the core pillars of its Second Poverty Reduction Strategy Paper. With significant resources allocated to the health sector, it is critical to measure whether or not the funding provided is reaching the target group that 22 The authors are with the World Bank. This work benefitted from the support of the Country Management Unit for Ghana. Liberia and Sierra Leone as well as the Human Development Network at the World Bank, as part of a broader study on faith and development in Ghana. The authors are grateful to Ishac Diwan and Rakesh Nangia for supporting this research. 121 needs it the most, namely the poor. Using benefit-incidence analysis, the objective of this paper is to measure to what extent public health spending is pro-poor. Previous studies on the benefit incidence of health spending in Ghana based on the third and fourth rounds of the GLSS surveys for 1991-1992 and 1998-1999 as well as other data have suggested that better off households tended to benefit the most from health spending (e.g. Demery et al 1995, Nyonator and Kutzin 1999, and Canagarajah and Xiao 2001: see also the analysis of Demographic and Health Surveys in Makinen et al 2011, and the qualitative fieldwork by Shojo et al 2012 as well as Wodon 2013). In this paper, we update the analysis using the fifth round of the GLSS, collected in 2005- 2006, and we test whether the benefit incidence analysis of public spending for health depends on the assumptions used to estimate unit costs. More precisely, using district- level financial data to compute district-level unit costs, this paper takes a closer look at the distribution of spending across both geography and income groups to understand if efforts to decentralize spending and invest in pro-poor services in health have succeeded in leading to a pro-poor benefit incidence. Another feature of this paper is that we compare benefit incidence results for both public health facilities and the facilities operated by the Christian Health Association of Ghana (CHAG) which also benefits from public funding, especially for staff salaries. Overall, the analysis suggests that public health spending remains regressive for both the public and faith-inspired sectors in large part because hospitals and clinics still benefit the better off more than the poor, independently of who operates the facilities. DATA AND METHODOLOGY Three sources of data are used in the paper. The first two sources are nationally representative household surveys. The first survey is the Ghana Living Standards Survey (GLSS5) implemented in 2005-2006. The GLSS is a multi-purpose household survey covering demography, health, education, employment, migration, housing, agriculture activities, non-farm self-employment, household expenditures, durable goods and, remittances and other incomes. The 2005/06 round of the survey was administrated to around 36,500 individuals grouped into 8700 households. This nationwide sample is deemed representative at the level of the ten regions. The second survey is the large sample (50,000 households) 2003 Core Welfare Indicator Questionnaire (CWIQ) survey, deemed representative at the district level. Both surveys distinguish between faith- inspired and other types of providers when asking about care sought by individuals, although this is done in slightly different ways in each of the two surveys. In the CWIQ survey, one question is asked as to whether each individual in the household has consulted a doctor, nurse, pharmacist, health professional, dentist or traditional healer for any reason during the last 4 weeks. For those who did consult, a subsequent question is asked as to the kind of health provider/facility the household member did see or visit. The modalities for the response are a private hospital/clinic, a public hospital/clinic, a community health center, a private doctor/dentist, a traditional healer, a missionary/hospital, a pharmacist/drugstore, and finally the option “Other�. Of those 122 providers, three will be considered as benefiting from public health spending: public hospitals/clinics, community health centers, and missionary/hospital facilities. Those data will then be combined with administrative data on public funding for health at the district level. For reasons explained below, in order to compute unit costs, public hospitals/clinics and missionary/hospital facilities are combined as one type of provider, while community health centers will be considered as a separate type of provider. In the GLSS5, the questions are asked differently, and more precisely. For those who consulted a health provider in the last two weeks, a question is asked as to where the consultation took place. 23 The modalities for the response are a hospital, a clinic, a MCH clinic, a maternity home, a pharmacy, a chemical store, a consultant’s home, the patient’s home, and “other�. The next question is whether this was a public or private facility, with three modalities for the question: public, private religious, and private non-religious. The combination of the two questions permits the identification of the faith-inspired as opposed to the public facilities, as well as the type of facility used. Here again, to compute unit costs, hospitals and clinics will be aggregated together, and treated separately from community health centers (MCH clinics). In addition to these two surveys, we use administrative data on public health spending. Budget data are available for all 138 districts, and the total budget included in the analysis is GHC 321,115,072. Five different budget lines are provided: salaries which account for 69.0 percent of the total budget (GHC 221,675,525), administration for the Ministry of health, accounting for 5.1 percent of the budget (GHC 17,244,558), administration for the NHIA accounting for 3.1 percent of the budget (GHC 9,824,628), services for the Ministry of health, with a budget share of 0.6 percent (GHC 1,861,186), and finally services for the NHIA with a budget share of 22.0 percent (GHC 70,509,167). We cannot differentiate in the budget data between funding allocated to faith-inspired as opposed to public facilities, and at the time of the budget data, it is likely that public facilities were better funded than faith-inspired facilities. But since then further steps have been taken to fund faith-inspired facilities more equally, and we simply assume here that all hospitals and clinics within a district receive the same funding proportionally to the number of visits to the facilities observed in the survey and the district allocation available in the data. We also assume that a visit to a community health center costs only half as much as a visit to a hospital or a clinic. We could change these assumptions, but it should not make much difference in the final results given that visits to community health centers account for only a small share of all visits to the three types of facilities combined (hospitals, clinics, community health centers). Given the above data, the methodology to conduct the benefit incidence analysis is standard. As noted among others by Castro-Leal et al (1999) and Demery (2003), the starting point in conducting a benefit incidence analysis lies in assessing the use of government services by households, which is done here by using the two nationally 23 A separate and useful question is also asked about whom was consulted for the most recent visit. The modalities for the responses are a doctor, a dentist, a nurse, a medical assistant, a midwife, a pharmacist, a drug/chemical seller, a traditional healer, a trained TBA, an untrained TBA, a spiritualist, and finally the option “other�. This question is however not used for the analysis presented in this paper. 123 representative household surveys mentioned above. This information is then combined with data on the cost for the government to provide the services, so that one can estimate the share of public spending that is allocated to different groups of households. In most analyses of benefit incidence analysis the data on unit cost is often provided only at a fairly aggregate level. Using the same notation as in Demery (2003), denote by Si the net government spending on health facilities of type i, with i =1,..,3 representing different types of facilities (hospitals, clinics, and health centers). The value Si should represent net costs for the government after having deducted fees and receipts from other cost recovery mechanisms, or net funding transfers as is the case of Ghana. Denote by Eij the number of patients using facilities from household group j (which may for example represent quintiles) in facility type i, and by Ei the total number of patients using this type of facility. Only students attending government-subsidized (i.e., typically public, but also in the case of Ghana faith-inspired) facilities should be taken into account in the estimation. Then, Si/Ei can be considered as the average unit cost for the government of providing health services in facilities of type i. The total implicit transfer in kind received from public health spending for households of type j, denoted by Xj, is then: 3 Si X j ≡ ∑ E ij (1) i =1 Ei Equation (1) makes it clear that the value of the benefits accruing to group j depends on both the unit cost of providing health services in different types of facilities, and the number of patients using the facilities in each household group. If S denotes the total public spending for health, the share of the total health subsidy accruing to group j, denoted by xj, is then given by: n 3 E ij S i n 3 xj ≡ ∑ ∑ = ∑ ∑ eij s i (2) k =1 i =1 E i S k =1 i =1 As pointed out by Demery (2003), a group’s benefit share of total public spending is equal to the weighted sum of the group’s share of total health visits for each type of facilities (eij), with the weight defined by the shares of the total budget allocated to the various types of facilities. In order to make comparisons between groups, or to assess whether the share of total benefits received by any given group can be considered as equitable, it is standard to compare the share of the benefits received by a group to the group’s population share in the total population. Consider now the case when public subsidies for health services vary across geographic areas, denoted by the subscript k (these represent districts in the case of Ghana). If Sik denotes the total cost of providing health services for facilities of type i in area k (k=1, …, n), and Eik is the number of patients using facilities of type i in area k, then, Sik/Eik is the average unit cost of public health services at level i in area k. These area-specific unit costs are important to factor in the analysis, because they may differ substantially between areas, and because in terms of outcomes, there is an implicit assumption (although not always verified empirically) that higher unit costs may be associated with higher quality in health services. In the case of health services in developing countries like Ghana, given that a large share of the costs are accounted for by the salary of health personnel (as well as drugs, but those are more often subject to cost recovery), the unit 124 costs may differ between areas due to both different rations of health personnel to patients (as well as occupancy rates for hospital beds for example), and differences in the qualification and thereby pay levels of health personnel (doctors, nurses, midwifes, etc.), which may also differ between areas, with poorer and more remote areas often staffed by less qualified personnel, but also possibly with lower occupancy rates or use of facilities by personnel. Denoting by Eijk the number of patients of group j using facilities of type i and area k, we have: n 3 S ik j ≡ ∑ ∑ E ijk XK (4) k =1 i =1 E ik As before, if S denotes the total public spending for health, the share of the total health subsidy accruing to group j, denoted by xj, is then given by: n 3E ijk S ik n 3 j ≡ ∑ ∑ xK = ∑ ∑ eijk s ik (5) k =1 i =1 E ik S k =1 i =1 A group’s benefit share of total public spending is still equal to the weighted sum of the group’s share of total visits to health facilities, but the sum is now taken for all types of facilities in each district or region, with the weight defined by the shares of the total budget allocated to the various types of facilities and districts or regions. As we shall see in the next section, the use of aggregate national-level unit costs as opposed to unit costs disaggregated by district or region can make a difference in the estimation of the shares accruing to each household group. UNIT COSTS The first step in the analysis consists in the estimation of unit costs by district. Differences in unit costs between districts are derived from the budgets allocated to each districts and the number of visits to the districts as computed in the surveys (using the appropriate expansion factors). Using the CWIQ data, we compute unit costs at both the district and the regional levels. With the GLSS5, we compute the unit costs only at the regional level, given the smaller sample size of the survey. In both cases, as shown in table 1, the unit costs vary widely across districts and regions, with higher unit costs observed in both the wealthiest (Greater Accra) and the poorest regions (Upper West). It could be that in Accra and a number of other comparatively better off regions, higher unit costs reflect the fact that health personnel attending to patients tend to be more qualified (most of the doctors tend to be concentrated in the wealthier parts of the country). By contrast, in poorer areas and due to affordability issues, the demand for care is lower (which translates for example in lower occupation rates for hospitals), which may also result in higher unit costs because of lower levels of utilization for the services made available. This then would also imply lower unit costs for districts and regions that tend to be in the middle of the welfare distribution. This “U� shape relationship between the incidence of poverty and average unit costs for visits to health facilities in a given district or region is clearly visible in Figure 1 (the size of the dots in the scatter plots are proportional to the size of the population in the various districts or regions in each Figure, 125 but not comparable for district as opposed to regional scatter plots; both linear and quadratic best fit lines using regressions weighted by population are provided). Table 1: Unit cost estimates per visit to a health facility Total cost Unit cost Unit Cost, CWIQ 2003 by region, GLSS 5 District District District in GHC 2005-2006 Average Minimum Maximum Western 22,994,468 5.6 6.9 2.9 25.5 Central 22,597,436 7.8 7.6 1.9 29.9 Grater Accra 60,990,935 14.9 19.9 9.2 23.5 Volta 27,849,206 11.5 9.3 4.8 21 Eastern 40,634,717 6.6 10.5 5.3 21.3 Ashanti 53,912,428 8.3 8.9 2.4 15.7 Brong Ahafo 40,507,241 11.6 13.8 4.3 30.9 Northern 24,579,028 5.8 9.6 2.9 35.2 Upper East 14,914,581 11.7 9.4 4.7 16.8 Upper West 12,135,021 12.5 19.8 13.5 27 National 321,115,061 8.9 10.7 1.9 35.2 Source: Authors’ estimation using CWIQ 2003, GLSS5 2005-2006, and administrative data. Figure 1: Relationship between unit costs and poverty levels at the regional level (a) CWIQ-based District Unit Costs and Poverty Headcounts 40.0 30.0 20.0 10.0 0.0 0.0 20.0 40.0 60.0 80.0 Poverty Headcount (in %) District-level Unit Cost (in GHC) Linear fit Quadratic fit Source: Authors’ estimation using CWIQ 2003, GLSS5 2005-2006, and administrative data. 126 Figure 1 (Continued): Relationship between unit costs and poverty levels at the regional level (b) CWIQ-based District Unit Costs and Poverty Headcounts 20.0 15.0 10.0 5.0 0.0 20.0 40.0 60.0 80.0 Poverty Headcount (in %) Region-level Unit Cost (in GHC) Linear fit Quadratic fit (c) GLSS5-based Regional Unit Costs and Poverty Headcounts 15.0 10.0 5.0 20.0 40.0 60.0 80.0 100.0 Poverty Headcount (in %) Region-level Unit Cost (in GHC) Linear fit Quadratic fit Source: Authors’ estimation using CWIQ 2003, GLSS5 2005-2006, and administrative data. 127 BENEFIT INCIDENCE The results of the benefit incidence analysis are provided in tables 2 through 4. Table 2 provides simply the share of visits to health facilities accounted by various population groups – namely the urban and rural populations, as well as the population ranked according to level of consumption per equivalent adult into five quintiles of well-being, from the poorest (Q1) to the richest (Q5) quintile. Table 3 then combines that information with the unit costs estimated in the previous section, using both the district-level and regional unit costs in the case of the CWIQ 2003 survey, and the regional level costs in the GLSS5. Finally table 4 repeats the exercise by considering national unit costs (thus no differences in estimation of the unit costs between areas). Consider first table 2 which is solely based on the survey data. A few findings stand out. As expected, visits to private facilities tend to be concentrated among the top quintiles of the distribution, and this is observed with both the CWIQ and the GLSS5 surveys. In the CWIQ survey, religious facilities tend to serve the poor significantly more than public facilities, but the reverse is observed in the GLSS5 survey. Thus, while in terms of the estimation of the unit costs the two surveys tended to generate similar results, in terms of the visits to health facilities, the results differ, so that it is difficult to reach a clear conclusion. The questionnaire of the GLSS5 is better to identify where individuals seek care than that of the CWIQ, but this is not a sufficient reason to discount the results obtained with the CWIQ. Said differently, a fair assessment of the results would be to say that on average, between the two surveys, the incidence of visits profile is not too different between the public and the faith-inspired providers. There is also a clear pattern observed in both surveys whereby visits to hospitals are more frequent in the top quintiles, while visits to community health centers are more frequent in the lower quintiles. Visits to clinics are more prevalent among the better off, but less so than visits to hospital. Given the cost of the services provided by various facilities and location effects (hospitals tend more often to be located in urban and therefore often wealthier areas), these results are as expected. 128 Table 2: Share of Patients Attending Facilities by Location and Quintile (%) Residence Area Welfare Quintile Total Urban Rural Q1 Q2 Q3 Q4 Q5 CWIQ 2003 All visits 47.1 52.9 13.2 16.5 19.0 22.1 29.2 100.0 Public facilities 44.0 56.0 14.6 17.7 19.4 21.2 27.1 100.0 Private facilities 61.1 38.9 6.7 12.5 17.7 25.3 37.7 100.0 Religious facilities 30.7 69.3 20.2 17.3 19.3 20.6 22.6 100.0 All hospitals/clinics 47.8 52.2 12.7 16.5 19.1 22.2 29.6 100.0 Public hospitals/clinics 44.9 55.1 14.0 17.7 19.5 21.4 27.5 100.0 Private hospitals/clinics 61.1 38.9 6.7 12.5 17.7 25.3 37.7 100.0 Religious hospitals/clinics 30.7 69.3 20.2 17.3 19.3 20.6 22.6 100.0 Community health centers 22.1 77.9 29.7 18.4 15.7 17.2 18.9 100.0 GLSS 2005-2006 All visits 43.2 56.8 11.0 15.8 19.1 22.5 31.6 100.0 Public facilities 42.0 58.0 12.3 16.9 19.5 22.7 28.6 100.0 Private facilities 47.9 52.1 7.9 13.3 16.0 22.4 40.4 100.0 Religious facilities 41.9 58.1 8.3 12.4 23.5 21.4 34.4 100.0 All hospitals 52.3 47.7 8.3 13.9 18.6 23.6 35.7 100.0 Public hospitals 50.1 49.9 8.9 14.9 18.9 24.0 33.3 100.0 Private hospitals 69.6 30.4 6.1 9.7 11.7 24.8 47.7 100.0 Religious hospitals 49.4 50.6 5.1 9.4 24.8 18.1 42.6 100.0 All clinics 33.1 66.9 13.8 18.9 19.7 20.9 26.7 100.0 Public clinics 29.8 70.2 16.9 21.1 21.0 20.5 20.6 100.0 Private clinics 39.4 60.6 8.2 15.5 17.0 20.4 39.0 100.0 Religious clinics 31.9 68.1 13.2 15.8 21.2 25.9 23.9 100.0 All community health centers 23.0 77.0 20.9 9.4 18.8 26.7 24.3 100.0 Public community health centers 17.9 82.1 27.9 7.8 10.8 23.9 29.6 100.0 Private community health centers 28.8 71.2 13.3 9.0 28.6 31.6 17.5 100.0 Religious community health centers 34.8 65.2 - 29.8 32.2 21.7 16.3 100.0 Source: Authors’ estimation. Note: no data on visits to religious community health centers in the bottom quintile. Table 3 provides the data not only in terms of visits, but in terms of benefit incidence, whereby the cost associated with the different facilities (which takes into account their location in various districts or regions) is also factored in. Note that this analysis includes only public and faith-inspired facilities, given that private facilities tend not to be funded by the state. By and large, taking into account the unit costs does not fundamentally change the results. This is in part because despite the “U� pattern mentioned earlier in the geographic distribution of unit costs, on average there are no linear patterns that differentiate unit costs between districts according to levels of poverty; this, together with the fact that there are poor populations in many different types of districts tends to average out the potential influence that differences in unit costs might yield (for example, more benefits due to higher unit costs would accrue to some of the households in the poorest parts of the country, but this is then compensated by the use of facilities by the poor in some of the areas that have medium levels of poverty, etc.) When the analysis is conducted using national level unit costs, the results (not shown here) are again similar. 129 Table 3: Benefit Incidence by Location and Quintile (%) Residence Area Welfare Quintile Total Urban Rural Q1 Q2 Q3 Q4 Q5 CWIQ 2003, District Level Costs All visits 54.5 45.5 12.8 15.9 19.2 21.7 30.3 100.0 Public facilities 55.8 44.2 12.2 15.8 19.2 21.9 31.0 100.0 Religious facilities 36.6 63.4 21.7 18.0 18.7 19.5 22.1 100.0 All hospitals/clinics 54.9 45.1 12.4 15.9 19.3 21.8 30.5 100.0 Public hospitals/clinics 56.3 43.7 11.7 15.8 19.4 22.0 31.2 100.0 Religious hospitals/clinics 36.6 63.4 21.7 18.0 18.7 19.5 22.1 100.0 CHC 31.4 68.6 34.3 16.2 12.8 15.8 20.8 100.0 CWIQ 2003, Regional Level Costs All visits 54.5 45.5 12.8 15.9 19.2 21.7 30.3 100.0 Public facilities 55.8 44.2 12.2 15.8 19.2 21.9 31.0 100.0 Religious facilities 36.6 63.4 21.7 18.0 18.7 19.5 22.1 100.0 All hospitals/clinics 54.9 45.1 12.4 15.9 19.3 21.8 30.5 100.0 Public hospitals/clinics 56.3 43.7 11.7 15.8 19.4 22.0 31.2 100.0 Religious hospitals/clinics 36.6 63.4 21.7 18.0 18.7 19.5 22.1 100.0 CHC 31.4 68.6 34.3 16.2 12.8 15.8 20.8 100.0 GLSS 2005-2006, Regional Level Costs All visits 46.1 53.9 11.8 16.6 19.6 22.0 29.9 100.0 Public facilities 45.9 54.1 12.2 17.1 19.4 22.2 29.1 100.0 Religious facilities 48.3 51.7 8.8 11.5 21.3 20.6 37.8 100.0 All hospitals 53.7 46.3 8.8 14.6 19.1 22.3 35.1 100.0 Public hospitals 53.5 46.5 9.3 15.2 18.8 22.7 34.0 100.0 Religious hospitals 56.0 44.0 4.5 8.1 22.7 18.2 46.5 100.0 All clinics 34.9 65.1 16.1 20.1 20.6 21.6 21.7 100.0 Public clinics 34.6 65.4 16.2 20.5 20.8 21.3 21.2 100.0 Religious clinics 37.7 62.3 15.0 15.9 19.2 23.8 26.1 100.0 All health centers 20.6 79.4 28.2 9.5 12.2 23.3 26.8 100.0 Public health centers 19.3 80.7 30.3 8.1 10.8 23.3 27.5 100.0 Religious centers 38.0 62.0 - 28.0 30.6 23.7 17.8 100.0 Source: Authors’ estimation. CONCLUSION The objective of this paper was to use two different but nationally representative household surveys to conduct a benefit incidence analysis of public spending for health. While many benefit incidence studies assume that unit costs are uniform with a country, we were able to use detailed district- and region-level unit costs to assess whether the average public cost of a consultation to a facility costs more in some districts or regions than in others. The analysis suggests substantial variation in transfers and unit costs between districts and regions, with higher unit costs in geographic areas with the lowest and highest poverty measures, and lower costs in-between – that is, unit costs reveal an inverted “U� relationship versus the incidence of poverty by district or region. The district- or regional level unit costs as well as the data on the use of facilities were then combined with the data on the socio-economic characteristics of households to assess whether public health spending reaches equally various segments of the population. Public health funding was found to be regressive overall, in large part because hospitals and clinics still benefited more in both surveys the better off than the 130 poor. Finally, the analysis also distinguished government or faith-inspired providers, and although there were differences in results for faith-inspired providers between the two surveys, overall the results suggest that public spending for faith-inspired providers does not necessarily reach the poor significantly better than public spending for public providers (in one survey, faith-inspired providers reached the poor more than public providers but it was the reverse in the other survey). REFERENCES Castro-Leal, F., J. Dayton, L. Demery, and K. Mehra. 1999. “Public Social Spending in Africa: Do the Poor Benefit?.� World Bank Research Observer 14(1): 49-72. Canagarajah, S., and X. Ye. 2001. “Health and Education Spending in Ghana in 1992- 98.� Policy Research Working Paper 2579, World Bank, Washington, DC. Demery, L., S. Chao, R. Bernier, and K. Mehra. 1995. “The Incidence of Social Spending in Ghana.� World Bank PSP Discussion Paper Series, Washington DC. Demery, L. 2003. “Benefit Incidence Analysis.� In F Bourguignon and L Pereira da Silva (eds.), The impact of economic policies on poverty and income distribution: evaluation techniques and tools. World Bank, Washington, DC. Makinen, M., S. Sealy, R. A. Bitran, S. Adjei, and R. Munoz. 2011. Private Health Sector Assessment in Ghana. World Bank, Washington, DC. Nyonator, F., and J. Kutzin. 1999. “Health for Some? The Effects of User Fees in the Volta Region of Ghana.� Health Policy and Planning 14 (4):329-41. Salisu, A., and V. Prinz. 2009. Healthcare in Ghana. Wien: Austrian Centre for Country of Origin and Asylum Research and Documentation, and Austrian Red Cross. Shojo, M., C. Tsimpo, and Q. Wodon. 2012. “Satisfaction with and Reasons for Choosing Faith-Inspired Health Care Provision in Ghana.� In J. Olivier and Q. Wodon, eds, The Comparative Nature of Faith-inspired Health Care Providers in sub-Saharan Africa, World Bank, Washington, DC. Wodon, Q. 2013. Faith and Human Development: Education and Health Services in Africa, New York: Palgrave MacMillan. 131 About this series... This series is produced by the Health, Nutrition, and Population Family (HNP) of the World Bank’s Human Development Network. The papers in this series aim to provide a vehicle for publishing preliminary and unpolished results on HNP topics to encourage discussion and debate. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations or to members of its Board of Executive Directors or the countries they represent. Citation and the use of material presented in this series should take into account this provisional character. For free copies of papers in this series please contact the individual authors whose name appears on the paper. Enquiries about the series and submissions should be made directly to the Editor Martin Lutalo (mlutalo@worldbank.org) or HNP Advisory Ser- vice (healthpop@worldbank.org, tel 202 473-2256, fax 202 522-3234). For more information, see also www.worldbank.org/hnppublications. The world bank 1818 H Street, NW Washington, DC USA 20433 Telephone: 202 473 1000 Facsimile: 202 477 6391 Internet: www.worldbank.org E-mail: feedback@worldbank.org