Community accountability at peripheral health facilities: a review of the empirical literature and development of a conceptual framework.

Public accountability has re-emerged as a top priority for health systems all over the world, and particularly in developing countries where governments have often failed to provide adequate public sector services for their citizens. One approach to strengthening public accountability is through direct involvement of clients, users or the general public in health delivery, here termed 'community accountability'. The potential benefits of community accountability, both as an end in itself and as a means of improving health services, have led to significant resources being invested by governments and non-governmental organizations. Data are now needed on the implementation and impact of these initiatives on the ground. A search of PubMed using a systematic approach, supplemented by a hand search of key websites, identified 21 papers from low- or middle-income countries describing at least one measure to enhance community accountability that was linked with peripheral facilities. Mechanisms covered included committees and groups (n = 19), public report cards (n = 1) and patients' rights charters (n = 1). In this paper we summarize the data presented in these papers, including impact, and factors influencing impact, and conclude by commenting on the methods used, and the issues they raise. We highlight that the international interest in community accountability mechanisms linked to peripheral facilities has not been matched by empirical data, and present a conceptual framework and a set of ideas that might contribute to future studies.

The majority of empirical papers we identified on community engagement at peripheral health facilities in low- or middle-income countries focused primarily on health facility committees and groups.

Key influences on the impact of the community engagement activities are: how committee and group members are selected and their motivation for involvement; the relationship between groups or committees, health workers and health managers; and provision of adequate resources and support by local and national governments.

International interest in community accountability mechanisms linked to peripheral facilities has not been matched by empirical data. We present a conceptual framework and a set of ideas for consideration in future studies.

Introduction

Public accountability has re-emerged as a top priority for health systems all over the world. Public accountability refers to ‘the spectrum of approaches, mechanisms and practices used by the stakeholders concerned with public services to ensure a desired level and type of performance’ (Paul 1991). Although there is a range of possible definitions of accountability, the essence is of answerability between sets of actors in relation to specific activities or interventions. Answerability can be enforced with either positive or negative sanctions, or internalized ethics such as codes of conduct (Standing 2004).

Direct engagement of the public, citizens or communities in accountability systems can be through ‘exit’ or ‘voice’ strategies (Paul 1991). Exit strategies relate to citizen ability to access alternative suppliers of a given service; voice strategies relate to their ability to influence the outcome of a service through some form of participation or articulation of protest or feedback. Voice strategies in the health sector can be through the ‘short route’ to accountability between service providers and users (e.g., through involvement of the public or communities in monitoring of services, strengthening consumer power to complain, or making the income of service providers dependent on accountability to users), or through the ‘long route’ to accountability between governments and citizens, which involves broader social and political change (Standing 2004). The use of either or both of these strategies, together with the more traditional accountability measures, determines overall levels of public accountability.

Accountability problems are found in health systems the world over, but there are particular problems in developing countries where governments have often failed to provide adequate public sector services for their citizens (Standing 2004). This concern has contributed to an increasing emphasis on direct public engagement in health delivery in developing countries. Strengthening community accountability is promoted as a right in itself, and to enhance quality of care, appropriateness of health service delivery for users, and patient satisfaction and utilization (Cornwall ; Standing 2004; World Bank 2004). Although there is a long history of community participation in the public health system, particularly in health education and disease prevention, the recent shift in health service delivery is ‘away from users as recipients of services designed for their benefit, towards communities being active makers and shapers of services, exercising their preferences as consumers and their rights as citizens’ (Cornwall ).

Mechanisms introduced to strengthen community accountability might be distinguished by expected impact (financial, performance, or political/democratic), and by depth of community involvement expected or achieved (Goetz and Gaventa 2001; Brinkerhoff 2004). Regarding depth of community involvement, frameworks distinguish between simple information giving to communities at one end of the spectrum, through consultation, to community influence and control at the other end (Arnstein 1969; Loewenson 2000). These frameworks highlight that the creation of opportunities for consultation do not in themselves lead to community influence and control, and that there can be an element of ‘manipulation’ or ‘tokenism’ in initiatives. They also point to the importance of issues of legitimacy, representation and health system–community relations in community accountability. Overall, community accountability arguably moves beyond community participation in requiring the health system to be responsive to the issues raised through participation. In this context, ‘responsiveness’ can be defined as changes made to the health system on the basis of ideas or concerns raised by, or with, community members through formally introduced decision-making mechanisms.

The potential intrinsic and instrumental benefits from strengthening community accountability in health delivery have led to significant resources being used by governments and non-governmental organizations to introduce and improve mechanisms. Some of these initiatives are linked to peripheral health facilities, which are important sources of care for low-income populations (Loewenson ). Mechanisms introduced include patient/citizen rights charters, suggestion boxes, health and development committees, health clubs and volunteers.

Given the international interest and potential importance of community accountability, but also possible negative consequences such as real or perceived manipulation of communities and therefore inappropriate use of scare health system funds for which there are competing demands, we reviewed the available empirical literature on mechanisms linked to peripheral health facilities. We draw on this literature to propose a conceptual framework for consideration in future research in this important area.

Methods

We searched the published literature for empirical papers on topics related to community involvement at peripheral health facilities in low- or middle-income countries [LMICs, as defined by Organisation for Economic Co-operation and Development (OECD)] and systematically documented the process. The search terms are shown in Table 1. We considered descriptive and evaluation papers focusing on urban or rural primary health care facilities (e.g., health centres, health posts, dispensaries, community pharmacies), where the authors described at least one measure to enhance community accountability that was linked with those facilities. We excluded papers published before 1990, in languages other than English, and where the setting was the hospital/district level. An initial PubMed search was conducted in 2006, and cross-checked and updated in late 2009. The papers identified were supplemented by hand searches of the following websites: World Bank, World Health Organization, Partnerships for Health Reform, id21 and TARSC (Training and Research Support Centre).

Table 1

Search terms used in searching electronic databases

Group Aa: Target population (Combined by ‘OR’)	Group Ba: Geographic location (Combined by ‘OR’)	Group Ca: Interventions (Combined by ‘OR’)
Health cent*	Third World countr*	Consumer participation
Dispensar*	Less-developed countr*	Consumer involvement
Clinic*	Sub-Saharan Africa*	Community action*
Primary health care	Low- and middle-income countr*	Public participation
Rural health facilit*	Developing countries [MeSH]	Client voice
Peripheral health facilit*	Africa South of the Sahara [MeSH]	Community accountability
Community pharmac*	Asia [MeSH]	Community participation
Health post*	Latin America [MeSH]	Community consultation
First level facilit*	Central America [MeSH]	Community representation
Primary level facilit*	South America [MeSH]	Local government
Rural health servic*	Poor resource countr*	Committee*
Health system	[List of all OECD-ranked LMICsb]	Patients charter
Health care		Community development forum*
Community health servic*		Community involvement
Community health center*		Social responsibil*
		Group process*
		Patient participation

a The three groups were ultimately combined with ‘AND’.

b Online at: http://www.oecd.org/dataoecd/35/9/2488552.pdf

MeSH is a controlled vocabulary produced by the National Library of Medicine used for indexing articles and books

Results

Papers screened and included

7825 papers were identified, 101 articles were selected for further consideration and 59 rejected following a review of the abstract. Seventeen of the remaining 42 papers were retained after reading the full text. The hand search of websites identified a further four unpublished documents.

Basic details of the 21 final papers are presented in Table 2. Half present data collected in sub-Saharan Africa. Other countries included are India, Colombia, Mexico, Cuba, Peru and Nepal. One paper includes comparisons across countries (Gilson ), and several compared findings between facilities or areas within a country (Loewenson 2000; Few et al. 2003; Jacobs and Price 2003; Kyaddondo and Whyte 2003; Loewenson ; London ; Mubyazi ). However, most papers focused on one or several health centres or villages in one area. A mixture of government initiatives and non-governmental organization (NGO) projects were included.

Table 2

Summary of empirical studies on community accountability and participation

Paper reference	Accountability mechanism(s); NGO/government initiative, and type	Geographical coverage	Type and location of health facilities	Data collection methods	Endpoints
Bishai et al. (2002)	Government Village development committees (VDCs)	Jumla and Nawal Parasi Districts, Nepal	Sub-health posts (rural)	Survey of randomly selected facilities Interviews with clinic administrators	Caste and gender characteristics of VDCs, and association of these characteristics with VDC financial contributions to facilities
Bjorkman and Svensson (2009)	NGO Citizen report cards; developed through meeting	9 districts in all four provinces of Uganda	Public dispensaries	Multi-method; pre–post surveys, as part of a randomized controlled trial including: Facility record review User and provider surveys	Quantity and quality of health care provision Health outcomes (child mortality and infant weight for age) Changes in all accountability steps along the way
Few et al. (2003)	NGO Health facility committees	Lusaka (Zambia) and Dar es Salaam (Tanzania)	Urban primary health facilities	Evaluation of two urban health projects, including: Document review Individual and group interviews with staff and users Survey of household members (including non-users)	Committee role in monitoring, planning and management Health-related activities undertaken by committee and other independent health-related activities Links to environmental health officers
Gilson and Erasmus (2006)	Government Ward committees Clinic committee	Local government area in Gauteng Province, South Africa	Clinics and community health centres (rural and urban)	Case studies (5 ward committees and 2 clinic committees), including: Document review Individual interviews with committee members, community representatives and health workers Group discussions with community members	Basic functioning of community accountability mechanisms Links with local government and communities Impacts on health care delivery (reported)
Gilson et al. (2001)	NGO in Kenya; government in Benin and Zambia Local structures linked to financing schemes—in Benin, Kenya and Zambia	18 communes in Benin, 12 Bamako Initiative sites in Kenya, and 8 districts in Zambia	Existing primary care facilities in Benin; extension to new community level pharmacies in Zambia and Kenya	Rapid appraisal including: Semi-structured interviews with health workers, committee members and district-level staff In Kenya, focus group discussion (FGD) with community members	Equity impacts of community financing activities, including: Involvement of community representatives in decision making about the fee system and its management
Iwami and Petchey (2002)	Government Health committees (CLAS committees)—operate as non-profit organizations	Peru	Health centres	Not clear (original document gives full details but no English version available)	In terms of citizen participation: awareness of CLAS operations and women's participation and leadership Links of presence of CLAS to user satisfaction and to user fees
Jacobs and Price (2003)	Government and NGO Health centre co-management committees (HCCMCs) and feedback committees (FBCs); Pagoda structure and associated volunteers	Two operational health districts in Cambodia	Rural health centres	Mixed methodology study comparing two approaches to community participation Observation Open-ended interviews with committee members Cross-sectional household surveys of women	Acceptability by committee members of their assigned duties Acceptance by women of committee members for stimulating participation in health-related issues. Depth using Rifkin's approach
Jacobs and Price (2006) and Jacobs et al. (2007)	NGO (District contracted out to an international NGO since 1999) HCCMCs and community-based organization (Pagoda structures)	Kirivong Operational Health District, Cambodia	Health centres (rural)	Questionnaires administered to equity fund beneficiaries and non-beneficiaries Document review (meeting minutes) FGDs with HCCMC chiefs	Contribution of equity funds to community participation Sustainability over 32 months; measured using elements and indicators linked to 7 components adapted from Sarriot et al. 2004a,b (e.g. sustainable health outcomes, organizational capacity, community capacity)
Khumalo (2001)	Government Patients’ rights charters Clinic committees Home-based care and support groups Community development forums	Bohlabelo district, Limpopo Province, South Africa	Primary health facilities, mainly clinics (rural)	4 case studies, involving: Structured interviews In-depth interviews FGDs Exit interviews	Awareness of and operational features of mechanisms and factors influencing functioning Influence of mechanisms on quality of care (focusing on TB) Impact of mechanisms on: encouraging community participation; ensuring explanation can be sought from health workers about decisions made; making health workers answerable; and offering authority in sanctions
Kyaddondo and Whyte (2003)	Government District health committees, sub-county health committees and health unit management committees (HUMCs)	Tororo and Busia districts, Uganda	3 health centres and 1 hospital (rural)	Document review and observation Interviews and FGDs with health unit workers and committee members	Health workers’ perceptions of relationship between health sector reform and health worker motivation
London et al. (2006)	Government Patients’ rights charters	National level, and Western Cape and Limpopo Provinces, South Africa	8 clinics, and 10 district and regional hospitals (rural and urban)	Qualitative rapid appraisal including interviews with: National, provincial and local authority actors Health managers in two provinces A range of stakeholders at nine facilities across two provinces	Attitudes, experiences and behaviours with regards to patients’ rights charters, and differences and similarities between groups Factors influencing the above
Loewenson et al. (2004)	Government Health centre committees (HCCS; committees initiated or revitalized by the Community Working Group on Health, which has civic organization membership)	Goromonzi, Makoni and Gweru Districts, Zimbabwe	HCCs (rural and urban)	Mixed methodology study comparing 4 wards with and 4 wards without HCCs, including: Cross-sectional surveys Key informant interviews FGDs using a participatory assessment tool Case studies on key emerging issues	Relationship between HCCs and facilities Representation of community interests in health planning and management at health centres Allocation of resources to health centre level, to community health and to preventive services Community access to and coverage by health interventions Improved quality of care
Meuwissen (2002)	Health centre committees	Tillaberi District, Niger	Rural health centres	Case study including: Literature review Data collected by health workers Observation of district activities Policy meetings	Socio-economic and socio-organizational problems that contributed to a well-formulated programme on cost recovery failing to perform
Mogensen and Ndulube (2001)	Government (with support from DANIDA) Facility committees	4 provinces in Zambia; 3 districts per province	Rural health facilities	Qualitative component of a wider health reform evaluation including: FGDs Individual in-depth interviews	The wider study looked at implementation of the reform. Of interest in this paper is how key stakeholders relate to one another, and how these relationships are impacted on by reforms
Mosquera et al. (2001)	Individual participation Customer service offices User associations	Cali district, Colombia	Primary level hospital/health centres (rural and urban)	Mixed methodology study Group discussions Structured and semi-structured interviews	The social representations of different actors (policymakers, civil servants, user associations and users) that may hinder or enable effective implementation of participatory policy
Mubyazi et al. (2007)	Village Development Committees Ward Development Committees	Lushoto and Muheza districts, Tanzania	Rural and urban community health centres	Exploratory case study Multi-stage sampling: 4 villages per district FGDs with committee members and households	Community knowledge of health sector report, their participation in health priority setting, and how committees perform their duties in relation to community expectation
Schmidt and Rifkin (1996)	Women groups Village health committees	Lushoto district, Tanzania	Rural health centre	Descriptive case study Group and key informant interviews Participant and non-participant observation Review of relevant documentsw	Breadth of community participation in needs assessment, leadership, organization, resource mobilization, and management
Sepehri and Pettigrew (1996)	Government and NGO Village health committees	Two villages linked to an NGO in Nepal	Community-run health post and state-run health post; both rural	Small scale descriptive study Questionnaire guided interviews Participant observation Personal interviews	Extent to which community financing widened scope and extent of participation
Uzochukwu et al. (2004)	Government District and village health committees	Oji River Local Government Area, Enugu State, Nigeria	Rural health centres	FGDs with committee members Participant observations of committee meetings Health facility in charge questionnaire	Knowledge of and satisfaction with committees Involvement of committees in the health centre Committee member and district staff awareness of and involvement in the Bamako Initiative
Zakus (1998)	Government Local (community) health committees	Oaxaca state, Mexico	Health posts and health centres (rural and peri-urban)	Interviews using structured questionnaires Participant observations	Examines, analyses and characterizes: Relationship between health system and elements of the external environment (including community participants) Status and opportunities for community participants (indicators of value, recognition and power including training, guidance, resources, ability to set targets)

Notes: CLAS: local committes for health administration; DANIDA: The Danish International Development Agency; HCCS: health centre committees

Sustainability over 32 months; measured using elements and indicators linked to 7 components adapted from Sarriot , (e.g. sustainable health outcomes, organizational capacity, community capacity)

Nine of the 21 papers presented both structured surveys and qualitative work. The rest were qualitative studies, typically combining document reviews with focus group discussions (FGDs), individual interviews and observations. Individual and group interviews involved those directly implementing the accountability mechanisms, and—less frequently—those expected to benefit from them (e.g., facility users or the wider community), or those expected to oversee them or respond to problems raised (e.g., supervisors or managers of health facilities). Papers have a range of ‘endpoints’ of interest (Table 2).

Several papers formally draw upon or feed into a theoretical framework, including Franco et al.'s framework for analysing the relations between health sector reform and health worker motivation (Franco ; Kyaddondo and Whyte 2003), Sarriot et al.'s Sustainability Assessment Framework (Sarriot ; Sarriot ; Jacobs ), agency and stewardship models as the basis for the role of the state in health (Iwami and Petchey 2002), and Cohen and Uphoff's framework on dimensions and contexts of participation (Sepehri and Pettigrew 1996). Zakus (1998) developed a theoretical framework based on dependency theory (which postulates that organizations will react to pressures in their external environment to secure the resources needed for survival), and Gilson and Erasmus (2006) tested some initial ideas about linkages between trust, accountability and health delivery. Loewensen present a conceptual framework including outcomes/impacts, proximal factors/functioning, and underlying factors of power and authority. Methodologically, the most heavily drawn upon framework is Rifkin et al.'s approach to describing and assessing community participation (see Figure 1) (Rifkin ).

Figure 1

Spider diagrams to describe and assess community participation

Note: In Rifkin's approach level of community involvement in each of five factors considered to influence the breadth or depth of community participation in a community health programme is considered by the assessment team on the basis of data collection and discussion. The five factors are needs assessment, leadership, organization, resource mobilization, and management. To show visually levels of community participation, a spider diagram is drawn (see Figure). For each factor, the breadth of community involvement is based on a consensus reached by the assessment team, with widest participation being marked furthest from the central point, and narrowest closest to the central point. The points on each line are then linked with each other, allowing the breadth of community participation to be visualized.

The accountability mechanism most frequently covered was committees (health centre and clinic committees, village health committees and ward committees), followed by community groups (in particular, women's groups). Only two studies covered other accountability mechanisms in detail; London (2006) on patients’ rights charters, and Bjorkman and Svensson (2009) on citizen report cards. Other mechanisms such as suggestion boxes and health clubs were sometimes mentioned in passing. We begin by reviewing data on committees and groups, and then turn briefly to the data on other mechanisms, before discussing the issues raised across all papers.

Committees and groups

Origin and intended role of groups and committees—intervention ‘content’

Most papers discuss committees or groups of some form, some health specific (Gilson ; Mogensen and Ngulube 2001), and others with a broader remit such as community development, with health being only one component (Gilson and Erasmus 2006; Mubyazi ).

Government initiatives tend to be implemented through bureaucratic mechanisms, should in theory receive regular programmatic support and resources, and are therefore often at least intended to be implemented on a large scale (Zakus 1998). Many papers discussing committees covered government-supported committees that were part of the routine health system setup, typically initiated as part of public service decentralization. Several papers deal with NGO-supported projects (Meuwissen 2002; Few ; Jacobs and Price 2006; Jacobs ). NGO initiatives often have more resources over a smaller area, and greater ability to engage existing community-based organizations.

While committee roles were not always clearly articulated in papers, health committees were often part of a wider set of interventions aimed at improving the effectiveness of health systems, quality of care and management of health centres, and community involvement in health. Roles typically incorporated:

promoting active participation in decision-making by community representatives, and

wider information and consultation with communities in general through discussion of issues relating to health facilities.

Committee membership often included a health worker representative, members of the local community, and appointed key figures. Eligibility and nomination/election processes were sometimes clearly specified. Two examples from Nigeria and Cambodia, respectively:

‘Committees consist of the person in charge of the health centre (who is the link between the government and the community), the primary school headmaster (who is the secretary), representatives of religious and women's groups/associations, representative of the Red Cross and Boys Scout organization, a representative of the town union, and some drawn from age group and pensioner's associations.’ (Uzochukwu , p. 158)

‘The recommended membership of the HCCMC [Health Centre Co-Management Committee] is three health centre staff, plus an elected community representative from each of the (an average of two) communities covered by the health centre … Membership of the FBCs [Feedback Committees] is larger, with the ‘ideal’ being the entire HCCMC plus a male and female elected representative from each village served by the health centre … members of both committees should be elected by the community they represent, with a mandatory 50% of village households required to turn out for such elections.’ (Jacobs and Price 2003, p. 401)

Most papers reported that committee members were trained in their roles, but as discussed in greater detail below, roles were often not clearly defined, and resources allocated to training were sometimes inadequate.

Reported achievements and impacts of committees and groups

Some impressive achievements were reported. In Cambodia for example, community participation was structured around Pagoda volunteers and involved managing an equity fund to provide user fee exemptions. The use of Pagoda volunteers was found to be a more effective and sustainable way of involving communities than establishing new community structures with formally elected representatives (Jacobs and Price 2003). Committee members were more accepting of their assigned duties in the Pagoda project areas, and mothers in villages with a Pagoda were more likely to disclose personal problems to committee members (75% vs 58%; P = 0.02). In a subsequent study the authors showed using surveys and spider diagrams that Pagoda volunteers were able to improve financial access for the poorest to public health services, and to increase the level of community participation in health. Working through the Pagoda structure was considered relatively sustainable and replicable (Jacobs ).

In Zimbabwe, four wards with Health Centre Committees (HCCs) performed better than four without, including in level of health resources within clinics, service coverage and community health indicators (Loewenson ). The association between HCCs and improved health outcomes was observed even in highly under-resourced communities and clinics. In Peru, local health committees based on grassroots self-help circles that developed during the 1980s reportedly identified unmet health needs at the local level, generated and effectively allocated resources to meet those needs, and developed a payment system that protected the poorest groups (Iwami and Petchey 2002). Regarding the latter, 25.1% of services of committee-run establishments were free at the point of use compared with 14.7% in sites run by the Ministry of Health.

Even in these successful cases, there were limitations to the achievements. For example, in Zimbabwe HCCs were not well known in the community, and although their functions included identifying priority health problems in the community, they did not have a direct influence over core budgets, and had little influence on how their clinics were managed and run (Loewenson ). With the Local Committees for Health Administration (CLAS) system in Peru, many committees were reportedly under-resourced, inadequately supported and poorly integrated with the national health system. In places, the system therefore remained fragile (Iwami and Petchey 2002).

In other settings, the impact was found to be at best more mixed, and at worst negative. For example:

In Zambia and Dar es Salaam, participatory observations suggested that committees were taking on an increasing role in monitoring, planning and management, and building up a sense of ownership (Few ). Users were reportedly voicing more complaints, in part due to improved awareness of their rights. However, the success of committees in prompting health-related activities in the communities by independent groups, as opposed to by themselves, was reportedly more mixed.

In Nigeria, interviewee comments suggested that community participation was enhanced through the Bamako Initiative, with committees being involved in health activities, the provision of equipment, and identifying those deserving exemption from fees (Uzochukwu ). However, committee members complained of exclusion from the co-management of user fees and revolving funds, and from priority setting or decision making.

Several studies suggested that mechanisms have not been thoroughly institutionalized. In Tanzania, Mubyazi described community participation as ‘rhetoric’ rather than ‘reality’. In Niger, Meuwissen (2002) reported that despite their initial enthusiasm, health centre committees failed to execute effective control over the administration of drugs and finance.

In Oaxaca state in Mexico, Zakus’ interviews and observations in over 40 health facilities revealed that auxiliaries and committee members were not given due recognition in terms of infrastructure, resources, support and training, and that they had no input into their communities’ health service targets and programmes. He concluded that the participatory structures and processes in that part of Mexico were ‘devoid of present purpose and future potential’, and that they had in fact become ‘additional dependencies of the health system …’ (Zakus 1998, p. 491). He reported that once community members were co-opted into the system, they were compromised into maintaining their obligations, although with little chance of performing them well. He notes that ‘… perhaps the greatest issue is that so much potential was being lost and systematically ignored’ (Zakus 1998, p. 491).

Factors influencing impact of groups and committees

Together the papers offer a rich range of almost entirely qualitative data on factors that influenced if and how committees and groups meet their intended objectives. These can be grouped into three inter-related areas: the selection, composition and functioning of groups and committees; relationships between committee members, health workers and health managers; and elements of the wider context.

1. Selection, composition and general functioning

Clearly a first step in the influence of groups and committees is whether or not they are successfully set up and whether they meet at all (Sepehri and Pettigrew 1996; Zakus 1998; Mubyazi ). In South Africa, qualitative work found that some wards had not established committees, despite it being a government health policy and health workers being supportive of the idea in interviews (Gilson and Erasmus 2006). Where committees had been established, there appeared to be a lack of consistent and regular functioning. For example there were problems with keeping to meeting schedules and having quorum, leading to malfunctioning and high numbers of dropouts. As one committee member described:

“… we have not been holding regular meetings, not attending some workshops like any other wards. In that note we ended up being torn apart as ward committee members. Now I think we were left with 4 or 5 persistent members in the ward committee.” (Gilson and Erasmus 2006, p. 17)

Part of the problem was linked to selection and composition of committee members. In practice, the selection of representatives was rarely democratic or transparent. In Niger for example, many committee members appeared from observation to be selected by the village chiefs, and the treasurer or ‘cash keeper’ was often seen to be closely related to other committee members or to the chief (Meuwissen 2002). The domination of males in committees, and of the relatively affluent or politically prominent or powerful, was also noted by others (Sepehri and Pettigrew 1996; Gilson ; Loewenson ).

Concerns with selection processes included that committee members had sought office for personal and political gain, that they did not represent the concerns of the wider community, or that they lacked leadership and administrative and technical skills. In Nepal for example, villagers were not only generally unaware of the existence of a committee, they also held differing views from health committee members about the quality of services provided by the village health centre (Sepehri and Pettigrew 1996). Problems of the link between committees and community members were also indicated in other studies (Gilson ; Few ; Loewenson ), with some reporting significant mistrust between the two groups (Mubyazi ). In Niger, committee members being closely related to each other led to difficulties in dealing with their misappropriation of funds (Meuwissen 2002).

The above problems contribute to and result from a range of factors reported qualitatively in studies, including an unclear mandate, reporting structure and legal position for many committees, lack of incentives for committee members, inaccessibility of some areas and difficulties of communication, and inadequate time and resources dedicated to training and supervision (Mogensen and Ngulube 2001; Iwami and Petchey 2002; Few ; Loewenson ; Gilson and Erasmus 2006).

Regarding clarity in role, there remained particular uncertainty, confusion and sometimes conflict regarding extent of decision-making power, which in practice was often limited. As an FGD participant in South Africa commented:

“… we did not think that we are supposed to do that [influence health care]. We do not have authority to do that. The government has not informed us to do that. Maybe occasionally, when we meet nurses in the community we would remind them that their job is to assist people who are sick. But it ends there.” (Khumalo 2001, p. 44)

Lack of incentives was highlighted in several papers. Meuwissen, for example, reported that committee members complained often about lack of financial rewards:

‘Committee members complained a lot about the lack of incentives for their time and investment. Being in need of money, handling thousands of [local currency] and having to understand that there is no money to be paid, appeared to be very difficult.’ (Meuwissen 2002, p. 307)

Possibly linked to a lack of incentives, in one study committee members were reportedly seeking exemptions for themselves and their families, leading to concerns among health workers that committee members were potentially reducing facility revenue (Khumalo 2001). Meuwissen (2002) highlights the centrality of the issue of motivation by suggesting that a fundamental question is ‘why should community members be motivated to commit themselves to the tedious task of regularly performing administrative supervision correctly over a long period of time’ (p. 312).

Building upon existing well-functioning community-based structures appeared to overcome some of the above challenges. For example, the success in rural Cambodia was largely attributed to the existing system of formally elected volunteers who were held in high esteem within their local communities (Jacobs and Price 2003; Jacobs ). Pagodas and associated volunteers appeared to offer the advantages of established (and apparently trusted) leadership, local organization, resource mobilization and management. These are all areas that need to be carefully built up in new organizations, a process that takes time and resources. Within the Pagodas the pivotal role of key authorities, and the importance of involving them in facility interactions, was also noted. In other settings or situations, community members may be selected from community groups which themselves have difficulties in negotiating and managing internal political, organizational and financial issues (Mogensen and Ngulube 2001). It is therefore by no means guaranteed that these groups have democratic decision-making processes, or that they have systems to incorporate or link to the more vulnerable groups.

2. Relationships between committees/groups, health workers and health management systems

Communities are never homogeneous, and initiatives almost always include one party encouraging another to take part. Community accountability is therefore inevitably embedded in social relations and power struggles (Mogensen and Ngulube 2001).

A key relationship for accountability mechanisms involving health facilities is between health workers and committee members. Realistically, committees cannot arise, be sustained and actively participate in health services without invitation and support from those that they deal with directly within the health care system (Mogensen and Ngulube 2001). However, imbalances in information and power often lead to health workers controlling committees (Gilson ; Mogensen and Ngulube 2001; Few ). As a result, committees can come to be seen as bodies designed to serve health centre objectives rather than community members (Few ). Alternatively, community members may come to see health care as the prerogative of the government, or doubt their ability to influence policy:

“[what] we often fear is that nurses will think that since we did not go to school we are talking rubbish, what will we do if we go to the clinic and these nurses ask us whether we understand the questions we are asking them? What are we going to say? We are uneducated; therefore we would not know how to answer any questions. Nurses might even say we went to the clinic to rule them …” (Khumalo 2001, p. 44)

Similarly, health workers may be unhappy with community supervision and management of facilities, seeing it as interference in their work by unqualified people (Mosquera ). This might relate to personal losses by health workers. Kyaddondo and Whyte (2003), for example, report that closer monitoring of health workers by committees and the necessity for health units to generate income led to health workers’ professional autonomy being infringed, and to loss of control over health care resources within the clinic (e.g., misappropriation of drugs, informal charging of patients and mismanagement of health unit funds) and outside (e.g., treating patients in their homes, running clinics and drug shops, selling medicine, and engaging in agriculture and trade). Ultimately, they reported that the social value and respect of the health worker was potentially affected by the presence of a committee. Such concerns may explain why health workers do not always want to open up financial information to the community (Uzochukwu ).

Mogensen and Ngulube (2001) argue that while the presence of users or their representatives has had positive consequences (e.g., health workers being disciplined and even sacked as a result of poor behaviour), government/community member relations have also been affected in unintended and sometimes damaging ways. Specifically, health workers are finding themselves in the centre of increasingly strained relations between the government and community members:

‘Health workers experience that they deliver services which are compensated or reciprocated neither by their employers (the government), since salaries are meagre and working conditions bad, nor from below, since patients’ contributions are not making a noticeable difference to the health workers’ living standards. In addition, users feel that due to the fees they pay (which do make a noticeable economic difference for them) they can make higher demands upon the health workers. The latter, however, only rarely have the resources with which to make any noticeable difference in quality of service. They therefore lose dignity in the eye of the ‘donors’ (the patients and the administrators who each pay them).’ (Mogensen and Ngulube 2001, p. 24)

The above shows the vital role that health managers potentially play in facility-based community accountability mechanisms (Mosquera ). How do district managers support and supervise selection of committee members and committee functioning? How high do managers prioritize responding to community issues and concerns in their range of activities? As Meuwissen suggests, another fundamental question to ask regarding why some mechanisms fail is:

‘why would district health team members be motivated to commit themselves to the [often] tedious task of administrative supervision and consider this as a normal part of their duties?’ (Meuwissen 2002, p. 312)

3. Broader contextual issues

Two important aspects identified as affecting the applicability and implementation of committees/groups were the government context and socio-cultural norms and priorities.

Regarding government control, of interest is the primary care level of the health system in which committees operate. Loewenson note that in Zimbabwe there is a virtuous cycle between the strength of the primary health care system and that of HCCs, with each positively reinforcing each other. Health worker outreach and reasonable investments in primary health care were noted to be important to trigger this virtuous cycle. In all settings, committee achievements will inevitably be compromised where they are given inadequate resources and power to make decisions in practice, or where local and national governments are inexperienced in responding to community members’ concerns (Mosquera ). An underlying issue is whether committees really are a response to a felt local need, or more a need to comply with policy directives on community participation that are enforced from above (Gilson and Erasmus 2006), often from outside the country. Zakus argues that community participation in the module programme in Mexico was implemented for its utility in supplying resources, rather than for democratic or intrinsic purposes, leading to major flaws in the participatory process and unimpressive health outcomes: From Niger, Meuwissen reported similarly that:

‘[the Ministry's centralized organizational structure] was insensitive to the important issues of community development. It lacked the desire to share power and let the communities make decisions, it failed to provide needed resources, and it lacked the capacity to embrace a multisectoral perspective.’ (Zakus 1998, p. 492)

‘the health service approach to community participation was, in effect, imposed – the communities were given the responsibility for the administration of the system but they had never asked for it, and did not take it on. Government officials advocated this approach as empowerment of the community, while it can be explained alternatively as a way to shirk their own responsibility.’ (Meuwissen 2002, p. 312)

Socio-culturally, Sepehri and Pettigrew (1996) argue that community participation is often conceptualized in western cultural terms of individualism and equality, while some societies are characterized by hierarchy, interdependence and action through personal relationships and social networks. They argue that in communities where people manipulate their multiple connections for access to resources, goods and services, where there are sharp divisions based on ethnicity, wealth, gender and power, and where treatment-seeking involves very contrasting ‘traditional’ and ‘modern’ health care, the applicability of community participation as envisaged through donors and governments can be called into question. In other communities, challenges may relate more to populations being atomized (Gilson and Erasmus 2006). In contrast, in some settings there are elements of socio-cultural norms and structures that appear to be highly suited to participatory mechanisms, such as the Pagoda system in Cambodia or the self-help circles in Peru. Regardless of the direction of influence, issues of politics and power within the wider community clearly play out in the functioning of accountability structures. As Mosquera et al. conclude:

‘… participation is not a matter of policies and legislation; it is a complex process involving belief, customs, ways of life and power.’ (Mosquera , p. 58)

Other accountability mechanisms

Of the few studies on other mechanisms that were identified, only two provided detailed empirical data: one on citizen report cards (Bjorkman and Svensson 2009), and one on patients’ rights charters (London ). Other papers provide additional information (McNamara 2006; London 2007), but are not included in Table 2 due to insufficient empirical data.

Report cards

Provider report cards compare providers within a specified geographic region on a routine basis according to certain standards of quality performance. In public reports, providers are named and performance data are presented to the public, potentially improving choice and ability to dialogue (McNamara 2006).

Bjorkman and Svensson (2009) present findings from a randomized field experiment in 50 communities from nine districts in Uganda with the primary objective of introducing community-based monitoring, including the use of citizen report cards. Local NGOs facilitated village and staff meetings in which community members discussed health service delivery relative to other providers and the government standard, and steps the providers should take to improve health service provision. A pre and post survey involving 5000 households, and records from 50 dispensaries, suggested the intervention was highly effective. There were improved health outcomes in intervention areas, including a 33% reduction in under-five mortality (144 vs 97 deaths), and improvements throughout the accountability chain (e.g., greater monitoring of providers by community members, shorter waiting times at facilities and less absenteeism). The authors attribute the success to overcoming two problems in community participation and oversight: lack of relevant information, and failure to agree on or co-ordinate expectations of what is reasonable to demand from the provider.

McNamara's (2006) overview of report cards draws on data from the Yellow Star Program implemented in 12 districts in Uganda. Facilities were evaluated quarterly using 35 indicators, and those receiving a 100% score for two consecutive quarters had a yellow star posted prominently on the facility. This intervention contributed to an improvement in average scores from 47% in the first quarter to 65% in the second, and to plans to scale up the programme nationwide. McNamara cautions, however, that challenges in all settings include: success being dependent on providers being able to respond to shortcomings, monitoring, providers potentially performing to the measures (i.e. focusing on those measured rather than other aspects of care), and inadequate government support.

Patients’ rights charters

Patients’ charters are guidelines that target the relationship between health professionals and users of health services, providing information on standards of care that patients can expect to receive and demand as a basic human right (London 2006).

In South Africa, London (2006) observed that patients’ rights posters and suggestion boxes were common in facilities, and that facility staff and managers were often positive about the charter and its ability to motivate staff. There were concerns about sustainability of positive effects however, and many negative sentiments and dynamics. Many providers, for example, complained that the charter gave patients rights without emphasizing their responsibilities, and led to patients ‘expecting miracles’ and to health workers being ‘taken for granted’, and that it failed to recognize health workers’ own stresses and challenges. Ultimately, the charter was seen as a threat or weapon, ‘a sword over the head of health workers’; increasing stresses and strains rather than improving relations.

Some concerns were attributed to the charter itself: in aiming to re-balance the power relations between providers and patients it is bound to bring about discomfort. Other concerns were the imposition of the intervention from above, inadequate inclusion of health providers with real ‘field’ experience in decision making, and inadequate support to local level understanding and adaptation of the charter. More broadly, in a context of wider health system and community resource constraints, a facility-focused charter was recognized to be unable to address infrastructure-related issues essential for improving quality of care.

In a later paper London touches again on patients’ rights charters in South Africa and Malawi and civic organization around health in Zimbabwe (London 2007). He highlights their potential in providing community members with a standard for negotiating quality of care with providers at their facilities. However, he points to challenges when charters are not developed with community input, and where they are introduced in a context where civic structures tend to exercise political patronage rather than play active roles on behalf of civil society.

Discussion

There is a renewed drive in international policy and practice to strengthen direct involvement of the public, citizens or users in health delivery. We conducted a review of empirical studies exploring the functioning and impact of mechanisms introduced to strengthen community voices in peripheral health facilities in developing countries. Our review was based on a systematic search strategy, though we note the limitation of including only papers in the English language, which may have led to the exclusion of some studies from Francophone and Lusophone Africa and from South America. Overall, we found that the international interest in this area has not been matched by empirical data. Research to date has focused on health facility committees and other groups, with limited information on other mechanisms.

Regarding the available empirical data, there are four methodological points worth highlighting. First, many papers use the term community ‘accountability’ in a way that overlaps with the broader concepts of community ‘engagement’, ‘involvement’ and ‘participation’. Awareness of this potential contributed to our inclusion of these broader terms in our literature search (Table 1). However, the lack of precision in the literature on definition of concepts can lead to challenges in comparing goals and effects across programmes. Secondly, relatively few studies presented good quality quantitative data using observable measures of impact. More often, authors drew on views and perceptions of the committees and community members, which typically differed by stakeholder group (i.e. between health workers and committee members, or committee members and community members). While this is expected given the different potential losses and gains by different parties, such differences may also relate to interview bias.

Thirdly, a new accountability mechanism is often introduced at the same time as other interventions with similar expected impacts, including other accountability mechanisms and user fee changes. For example, Few point out that while the greater sense of ownership of health facilities may reflect the creation of the facility committees, it may also relate to the payment of fees through cost-sharing schemes. There are therefore numerous confounders which are difficult to account for without control groups. Fourth, the most widely used framework for assessing depth of community involvement is Rifkin et al.'s spider diagram (Figure 1) (Rifkin ). This is a valuable tool given the differences of opinions and views among key stakeholders, and the importance of understanding depth of involvement. However, a challenge is that these diagrams are constructed based on the consensus reached by an assessment team, with relatively little information presented in papers on what data were collected to feed into assessment teams’ deliberations, and how differences in opinion and group dynamics among the assessment team itself were handled.

The above caveats aside, the empirical literature suggests that while there is significant potential for community involvement in peripheral facilities, there is also a range of challenges to meeting this potential, including:

Complexities around defining ‘communities’ and their ‘representatives’: individuals are often members of a range of different communities, communities are far from homogenous (differing, for example, in age, gender, ethnicity and socio-economic status), and mechanisms for selecting individuals to represent those communities can be highly problematic. While existing community structures can be usefully engaged, these structures are themselves not necessarily democratic in origin or in decision making.

The functioning of community accountability mechanisms: particular challenges include members’ lack of clarity in roles and responsibilities, avoiding politicization, dilemmas related to voluntary participation/remuneration, difficulties in identifying the extent to which views are held by different community members, information and resource asymmetries between health staff and community representatives, and building trustful relationships in these contexts.

Views on if, when and how to involve communities can differ significantly between stakeholders, often leading to limited depth of accountability. Within communities challenges may include lack of awareness of rights, responsibilities and representatives, and limited ability to engage with health providers. Within health systems there may not be strong perceived value given to community input, accountability systems may discourage external influence, and there may be inadequate mechanisms and motivation for co-ordination and supervision. Community and heath system issues are all affected by the wider social, political and cultural environment. Factors such as availability of democratic fora, focus on human rights and availability of information can be key.

Given the paucity of published empirical data concerning community accountability in health delivery in general (i.e. beyond peripheral health facilities), and the prominence it has gained in policy and practice, many authors have called for further research (Zakus 1998; Cornwall ; Gilson ; Standing 2004; World Bank 2004; Cornwall and Nyamu-Musembi 2005; Mubyazi ). Considering the undeniable importance of qualitative research in this field, but also the potential contribution of quantitative data, we propose mixed methodology experimental studies wherever possible. Such studies will need to take into account the challenges suggested above and those outlined by Abelson and Gauvin (2006), including:

Difficulties in defining end-points: impacts can take years and may be difficult to disentangle from other events;

Public participation activities may be well run according to some criteria but not others; and

The problems of perceived vs actual impact, and the shortage of properly tested measurement tools.

On the basis of the above review, and drawing on policy analysis frameworks (Gilson ; Gilson and Raphaely 2008; Walt ; Buse ), future research would benefit from carefully considering four inter-related areas (Figure 2):

Figure 2

Factors influencing the functioning and impact of community accountability mechanisms

Content: What is the design of the accountability mechanism and how is it supposed to work? Which aspects of service delivery are expected to be impacted upon and how? What is the depth of community involvement planned at each stage of a policy or programme?

Process: How is the accountability mechanism actually working? How is the mechanism incorporated in practice into the health system hierarchy?

Context: What are the wider contextual issues—at the health system, national, international and community levels—that might influence the above areas? Particularly key for community accountability mechanisms linked to health facilities are likely to be health system factors including functioning of the primary care system, level and form of decentralization, and other forms of accountability in place, all of which can influence decision-making space to respond to community demands (Bossert 1998; Bossert and Beauvais 2002).

Impact: The above factors are together likely to influence impact, or ‘responsiveness’ of the health facility or health system to ideas or concerns formally raised by community members or their representatives. Although the ultimate goal of responding to community views is to improve health outcomes, measuring this impact is beyond the scope of many studies.

Ideally, future studies would include both process and outcome evaluation data. Given the centrality of depth of community involvement, both as a goal in itself and in being key to responsiveness, the measurement of this concept requires special consideration. We suggest drawing on Rifkin's spider diagrams and on the tables presented in Murthy and Klugman's paper (Murthy and Klugman 2004) to design tools which assist in the collection of data to be discussed by different stakeholders, including the assessment team. Murthy and Klugman's tables, and earlier work by others including, for example, Arnstein (1969), highlight the different potential degrees of participation that may underpin interventions, and how these might influence the definition of community, rationale for community participation, and scope and mode of participation employed.

Conclusion

Strengthening direct involvement of the public, citizens or users in health delivery should be supported by carefully designed empirical studies. Relatively little empirical information is available currently regarding mechanisms linked to peripheral facilities. The available data suggest that such initiatives will face challenges in relation to defining communities, in day to day functioning, and with regards to support at community, facility and at higher levels. Nevertheless, recent studies illustrate the potential of such interventions, and study designs that are valuable in contributing to the evidence base. We have presented a framework in this paper which we feel would be valuable in future studies on related topics.

Funding

All of the authors are members of the KEMRI-Wellcome Trust Research Programme in Kenya (core grant WT077092). SM is also supported by a fellowship from the Wellcome Trust (WT 085418), and CG is a member of the Consortium for Research on Equitable Health Systems, which is funded by the UK Department for International Development. This paper is published with the permission of the Director, Kenya Medical Research Institute (KEMRI).

Conflict of interest

None declared.