Implementing large-scale programmes to optimise the health workforce in low- and middle-income settings: a multicountry case study synthesis

Transcription

1 Tropical Medicine and International Health doi: /tmi volume 19 no 12 pp december 2014 Implementing large-scale programmes to optimise the health workforce in low- and middle-income settings: a multicountry case study synthesis Unni Gopinathan 1, Simon Lewin 1,2 and Claire Glenton 1 1 Norwegian Knowledge Centre for the Health Services, Oslo, Norway 2 Health Systems Research Unit, Medical Research Council of South Africa, Cape Town, South Africa Abstract objectives To identify factors affecting the implementation of large-scale programmes to optimise the health workforce in low- and middle-income countries. methods We conducted a multicountry case study synthesis. Eligible programmes were identified through consultation with experts and using Internet searches. s were selected purposively to match the inclusion criteria. documents were gathered via Google Scholar and PubMed and from key informants. The SURE Framework a comprehensive list of factors that may influence the implementation of health system interventions was used to organise the data. Thematic analysis was used to identify the key issues that emerged from the case studies. results s from Brazil, Ethiopia, India, Iran, Malawi, Venezuela and Zimbabwe were selected. Key system-level factors affecting the implementation of the programmes were related to health worker training and continuing education, management and programme support structures, the organisation and delivery of services, community participation, and the sociopolitical environment. conclusions Existing weaknesses in health systems may undermine the implementation of largescale programmes to optimise the health workforce. Changes in the roles and responsibilities of cadres may also, in turn, impact the health system throughout. keywords health systems, human resources for health, task shifting, implementation, governance, service delivery Introduction According to the 2006 World Health Report (WHO 2006), the global shortage of approximately 4.3 million health predominantly affects countries in sub- Saharan Africa, Asia and Latin America. One strategy that has been used to address these shortages is to optimise the existing workforce. Task shifting also known as task sharing (Lehmann et al. 2009) is one approach to optimising the workforce. Task shifting is defined as the training of cadres who do not normally have competencies for specific tasks to deliver these tasks and thereby increase levels of healthcare access (WHO 2012). The delegation of tasks from doctors to nurses (Lehmann et al. 2009), from doctors to other mid-level cadres (Brown et al. 2011) and from nurses to other mid-levels cadres (Mullan & Frehywot 2007) are examples of task shifting. The training of lay or community health to deliver basic health services is another form of optimisation (Lewin et al. 2010). While lay health do not necessarily take on tasks normally delivered by higher level health, they often support these health in other ways and also deliver promotional, preventative or other healthcare services that would otherwise have been neglected. Optimisation of the health workforce, including task shifting, has been proposed or used for a wide range of health issues, including HIV care (Callaghan et al. 2010; Fairall et al. 2012), maternal health (Gessessew et al. 2011; Nabudere et al. 2011) and family planning (Hoke et al. 2012). How large-scale programmes to optimise the health workforce in low- and middle-income countries (LMICs) should best be implemented, however, remains unclear (Fulton et al. 2011). This synthesis forms part of a series of studies and systematic reviews intended to inform WHO s Recommendations for Optimizing Health Worker Roles to Improve Access to key Maternal and Newborn Health Interven John Wiley & Sons Ltd 1437

2 tions through Task Shifting (OPTIMIZEMNH) (WHO 2012). As part of this work, a series of systematic reviews (Colvin et al. 2013; Glenton et al. 2013a,b; Rashidian et al. 2013) was conducted to explore the factors affecting the implementation of initiatives to optimise the health workforce. These reviews identified a large number of studies but had two important limitations. Firstly, most of the studies were of relatively small-scale or pilot programmes and focused on factors unfolding at the level of programme delivery in communities and primary care facilities. Secondly, we anticipated that, compared to small-scale pilot programmes, the implementation of large-scale programmes to optimise the health workforce would be associated with different issues related to programme functioning. Large-scale programmes, for instance, are often initiated by national governments and require the support of their national health systems, including access to training institutions, infrastructure, supply chains and referral systems. Compared to the flexibility of small-scale implementation, large-scale programmes implemented over a longer period of time may also require a higher degree of institutionalisation and standardisation. We decided therefore to undertake a multicountry case study synthesis focusing on factors affecting the implementation of large-scale programmes to optimise the health workforce. By studying these large-scale programmes, we hoped to reveal a different range of implementation factors and experiences and particularly upstream system-level factors associated with programme policies, governance, financing, planning, management and organisation to those already identified in the systematic reviews that had informed the OPTIMIZ- EMNH guidance (WHO 2012). This synthesis, we anticipated, would expand our understanding of the opportunities and constraints of optimising the health workforce at scale. Our objective was to identify factors affecting the implementation of large-scale programmes to optimise the health workforce in LMICs. Methods We conducted a synthesis of evaluations and studies of large-scale programmes designed to optimise the health workforce in LMICs. To address our research objective, a case study approach was used in which naturally occurring cases of health workforce optimisation at scale (as opposed to changes to the health workforce made as part of research projects or experiments) were examined (Hammersley 1992). The case study approach involves analysing phenomena in real life settings using a range of different types of evidence. This approach is particularly useful when it is important to understand a phenomenon in relation to its context (Yin 2009) and to compare and explore processes within and across programmes (Mitchell 1983; Patton 2002). selection Potentially eligible programmes were identified through consultation with experts and through Internet searches. We aimed to select programmes that: were focused on primary healthcare service delivery for maternal and child health, were located in Africa, Asia and South America, covered a broad variety of cadres, such as lay health, nurses, midwifes and other mid-level providers (cadre definitions used in this study are based on those included in the OPTIMIZEMNH recommendations) (WHO 2012, 2013), had documentation available in English, operated at a national (or very large) scale and had been running for at least 5 years (programmes that were no longer functioning also had to have run for a minimum of 5 years) We were also aware that the number of programmes included in our synthesis would need to be limited because of the resource-intensive nature of the data collection and analysis. Seven programmes were selected: (i) the Health Services Extension, Ethiopia; (ii) the Emergency Human Resources (EHRP), Malawi; (iii) the National Village Health Worker, Zimbabwe; (iv) the National Rural Health Mission, India; (v) the Primary Health Care, Iran; (vi) the Family Health, Brazil; (vii) and the Medicina Simplificada Program, Venezuela. Table 1 summarises the key features of these programmes. These included programmes did not always explicitly involve the shifting of tasks from one health worker cadre to another. In several programmes, optimisation focused on trying to ensure that previously neglected activities were incorporated into the tasks of particular health worker cadres. Although a large number of health workforce optimisation initiatives were identified, few operated at a national scale, had a sufficient level of documentation in English, or covered a range of cadres. Those programmes that were considered, but not included, are listed in Table S1. Data collection, analysis and synthesis Evaluation reports, programme guidelines and published studies were gathered for each selected programme using John Wiley & Sons Ltd

3 Table 1 Key characteristics of the selected programmes* Characteristics Country and name information Ethiopia: Health Services Extension Malawi: Emergency Human Resources Zimbabwe: National Village Health Worker India: National Rural Health Mission Iran: Primary Health Care Brazil: Family Health Venezuela: Auxiliares Medicine Simplificada Year launched Still functioning? Yes No No Yes Yes Yes Yes Broad description of role in improving maternal and child health Improve equitable access to preventive and essential health interventions through community-based health services, targeting households and in particular women, mothers and children at the village level Improve the acute shortage of health in the public health sector, and through strengthening of human resources for health facilitate implementation and delivery of interventions in the Essential Health Package, which includes priority interventions for maternal and child health Improve access to essential basic health services to the population by ensuring that each community has a grassroot health worker available Improve access to health care for rural people, in particular the poor, women and children Initial scope of the programme was infectious disease and maternal and child health, which now has been expanded to elders health, youth health, noncommunicable diseases and health promotion. Restructure the delivery of primary health care, including key intervention for maternal and child health and preventive services, through family health teams working at the community level. Improve provision of basic health services as well as health education and preventive services to the country s rural population 2014 John Wiley & Sons Ltd 1439

4 Table 1 (Continued) Characteristics information Country and name Ethiopia: Health Services Extension Types of health in the programme 1. Voluntary lay health 2. Lay health (Health Extension Workers) 3. Mid-level providers (Health Officers) Renumeration The Health Extension Workers and the Health Officers are salaried ; voluntary lay health receive non-financial incentives Malawi: Emergency Human Resources Scale-up of 11 priority cadres, including medical assistants, clinical officers, physicians, nurses and midwives. Lay health (Health Surveillance Assistants) were not among the priority cadres, but have been scaled-up during the course of the programme. All cadres are salaried Zimbabwe: National Village Health Worker 1. Lay health (Village Health Workers) 2. Medical assistants 3. Health assistants All cadres are salaried. Village Health Workers also received nonfinancial incentives India: National Rural Health Mission 1. Lay health (Accredited Social Health Activists ASHAs) 2. Mid-level providers (Auxiliary Nurse Midwives) ASHAs receive performancebased compensation; the ANMs are salaried Iran: Primary Health Care 1. Lay health (Behvarzes) The behvarzes are salaried Brazil: Family Health Health in the Family Health Teams 1. Lay health (Community Health Agents) 2. Assistant nurses 3. Nurses 4. Physicians All cadres are salaried Venezuela: Auxiliares Medicine Simplificada 1. Lay health (Medicine Simplificada Auxiliaries) The lay health are salaried John Wiley & Sons Ltd

5 Table 1 (Continued) Characteristics Country and name information Ethiopia: Health Services Extension Malawi: Emergency Human Resources Zimbabwe: National Village Health Worker India: National Rural Health Mission Iran: Primary Health Care Brazil: Family Health Venezuela: Auxiliares Medicine Simplificada Management structure and allocated management functions Woreda Health Offices and Woreda Health Councils are responsible for coordinating and implementing the programme at the local level. Functions include employment and deployment of Health Extension Workers, allocation and management of resources, coordination of in-service training, implementation of national guidelines, procurement of medical supplies and other tasks District offices have responsibility for human resource management, including personnel policy, performance management, training, human resource data systems, human resource strategy development and leadership and management District Council and District Health Teams were responsible for coordinating the programme at the local level District Management Units and the Office of the Chief Medical and Health Officer coordinate the programme activities at the local level. Tasks include the transfer of funds to the district and village level, and monitoring the implementation of policies and results in the district District Health Centers supervise and coordinate the activities of rural health centres. They organise health education activities and most of these centers are attached to specialised training centers responsible for the selection, training, deployment, supervision and monitoring of the behvarzes Municipal Health Departments are responsible for the provision of institutional support to family health teams. These Departments are also responsible for supporting facilities during the expansion and implementation of the Family Health Strategy, and for the development of training and education institutions for health in the family health teams Regional Health Service Administration. Specific functions unknown *Data sources for each programme are available in Table S2. The Emergency Human Resources was a time-limited 6-year programme to address the immediate human resources for health crisis in Malawi, by focusing on retention, deployment, recruitment, training and tutor incentives for 11 priority cadres. The Zimbabwean Village Health Worker programme declined in the 1990s and almost ceased to function during this period. Efforts to revitalise the programme were initiated in (Ministry of Health & Child Welfare 2010; Todd et al. 2010) John Wiley & Sons Ltd 1441

6 searching Google Scholar and PubMed. At least one key informant who had worked with or evaluated each programme was identified through the networks of the research team and contacted to identify further documentation (Table S2). The conceptual framework of this study was based on the SURE (Supporting the Use of Research Evidence) framework. This framework, which provides a comprehensive list of possible factors that may influence the implementation of health system interventions (The SURE Collaboration 2011), was used to develop our data extraction sheet and inform our analysis (Table 2 and Table S3). For each country case study, a narrative was prepared which included a timeline of key events. One author (UG) read each of the included documents, extracted data from them, and made summaries of these data. UG, assisted by the two other authors, read and re-read these data summaries and identified key themes. The definitions and boundaries of each emerging theme, and how these themes related to the SURE framework, were discussed by all three authors. Similar themes that emerged from different country case studies were categorised together within the relevant SURE framework category. These categories were then further discussed and adapted to ensure that they captured the key themes adequately. The Table 2 Key domains of the SURE Framework for the identification of factors affecting the implementation of policy options (adapted from the SURE Framework) (The SURE Collaboration 2011) Level Recipients of care Providers of care Other stakeholders (including other healthcare providers, community health committees, community leaders, programme managers, donors, policymakers and opinion leaders) Health system constraints Social and political constraints Factors affecting implementation Knowledge and skills Attitudes regarding programme acceptability, appropriateness and credibility Motivation to change or adopt new behaviour Knowledge and skills Attitudes regarding programme acceptability, appropriateness and credibility Motivation to change or adopt new behaviour Knowledge and skills Attitudes regarding programme acceptability, appropriateness and credibility Motivation to change or adopt new behaviour Accessibility of care Financial resources Human resources Educational and training system, including recruitment and selection Clinical supervision, support structures and guidelines Internal communication External communication Allocation of authority Accountability Community participation Management and/or leadership Information systems Scale of private sector care Facilities Patient flow processes Procurement and distribution systems Incentives Bureaucracy Relationship with norms and standards Ideology Governance Short-term thinking Contracts Legislation or regulation Donor policies Influential people Corruption Political stability and commitment John Wiley & Sons Ltd

7 categories reported here therefore do not match exactly those in the SURE framework (Table 2 and Table S3). In this analysis, we were interested primarily in factors affecting optimisation at scale for the selected programmes rather than in the effectiveness of these programmes. To assess effectiveness would have required a very different type of data, which this synthesis was not designed to identify or assess. The study s initial findings were presented, via , to the key informants and their feedback informed further analysis. Finally, the themes within each category were summarised in a summary of qualitative findings table (Table S4). Appraisal of study quality A wide range of reports and studies were examined. Some studies did not explicitly describe their data collection methods. In others, several qualitative and quantitative methodologies were applied. This made it impossible to use a single quality assessment tool. We decided therefore not to assess the methodological quality of each contributing report or study, or to evaluate the certainty of the evidence for each finding. Instead, an overall assessment of the strengths and weaknesses of the included data is provided (Box 1). Box 1 Methodological strengths and limitations of the included case study material Limitations Our synthesis only included material published in English and the range of programmes eligible for inclusion was therefore restricted. s from non-english speaking countries were considered but only if the evaluations were available in English. This may have limited the amount of data available to us, particularly in the case of the Venezuelan and Brazilian programmes. Some reports lacked adequate methodological descriptions of the data collection and analysis. The limitations of the programme data included: Sparse data for some programmes for which very few reports could be identified. Most reports covered specific periods in the programmes and this made it difficult to create a coherent understanding of them over their entire lifecycle. The amount of available contextual information about the programmes varied. Information about system issues such as the regulation of cadres, the policies that may have shaped the practice and organisation of health care, and the financing mechanisms for these programmes was particularly limited. Some of the reports were of large-scale evaluations of the programmes as a whole and provided useful, broad assessments. Others were studies of small components and these provided in-depth information. Reconciling these data was not always a straightforward process. Many reports of national or provincial programmes focused on small geographic areas. It was not always clear whether the findings were generalisable to the programmes as a whole. Limited data were found on cadres other than lay health. Reports tended to focus on the challenges facing programmes rather than processes that worked well. Information about good programme practices was limited. The reports we identified had not always been undertaken with the purpose of evaluating the implementation of the programmes themselves or of optimisation activities specifically. The information on questions relevant to the synthesis was therefore not always adequate. Strengths The use of a conceptual framework strengthened the categorisation and analysis of factors affecting the implementation of health programmes. Including evaluation reports from the grey literature meant that more information was available to address questions regarding upstream factors that affected the implementation of the programmes. The varied material identified for each programme (including progress reports, external evaluations and in-depth qualitative studies conducted at different points in the programme lifespan) allowed for the triangulation of sources, methods and timeframes. We attempted to ensure that a person familiar with each programme and setting reviewed the analysis. This meant that weaknesses and gaps in the data could be addressed John Wiley & Sons Ltd 1443

8 Results All selected programmes (Table 1) were implemented at a national scale, lasted at least 5 years or more, and collectively covered a broad range of geographic settings. Our goal was to identify programmes which included a wide variety of cadres. However, almost all the programmes we identified focused on the use of lay health even when other cadres were included. The study data were drawn mostly from programme evaluations (such as those commissioned by Ministries of Health) and from studies conducted by academic institutions or external agencies (including papers published in peer-reviewed journals). Our evaluation of two of the case studies also drew evaluation material that was relatively old. Partly, this was because the programmes were implemented far earlier in these instances Venezuela in the 1960s and Zimbabwe in the 1980s and therefore had long histories in the field. The Zimbabwean programme largely ceased to function in the 1990s and so little contemporary data are available (Ministry of Health & Child Welfare 2010; Todd et al. 2010). This case study was included as it is an example of a programme that encountered significant problems. We turn now to the key cross-country synthesis findings about the factors affecting the implementation of these large-scale programmes. These findings are also summarised in Table S4. Health worker training and continuing education Strategies for increasing enrolment and facilitating scale-up of training are important for programme implementation. Increasing the number of available health was an important objective of these programmes, and a number of strategies were adopted to increase the enrolment of student cadres and ensure that health received sufficient training. These strategies included collaboration with other ministries, as well as financial, educational and other incentives directed towards potential students. The rapid expansion of Ethiopia s Health Services Extension was enabled by a strategy in which theoretical training was offered through training institutions run by the Ministry of Education, and practical training was offered at health centres (Kitaw et al. 2007; Bilal et al. 2011). However, concern was expressed about whether training large numbers of health might be beyond the teaching capacity of the training institutions concerned (this issue is discussed in the section Human resource shortages affect training and supervision ) (Bilal et al. 2011). In Malawi, student enrolment rose significantly from 2004 after the government subsidised enrolment fees for various health professional training institutions as part of the EHRP. This increase in enrolment was partly facilitated by practical measures at training institutions such as the provision of off-campus accommodation and better use of the space available for training (Management Sciences for Health 2010). However, when the subsidy for training was cut by the government in 2009, enrolment levels dropped. In Iran, policymakers shifted the training of lay health (a cadre known in Iran as behvarz) to universities in 1999 (Javanparast et al. 2011a). The stated intention was to enrol more rural high school graduates (Javanparast et al. 2011a), develop a better educated group of behvarzes and thereby improve the quality of care provided by this cadre. However, we were unable to locate evidence on the effects of this policy change on enrolment levels. Human resource shortages affect training and supervision. Several programmes highlighted the fact that training was affected by a lack of trainers. In India, for example, a 2011 evaluation of the ASHA (the lay health worker component of the National Rural Health Mission) claimed that the delivery of the training had been slowed by limited human resources (National Health Systems Resource Centre 2011). Similarly, an assessment of the Health Services Extension in Ethiopia in 2007 concluded that the number of trainers for the first enrolment was inadequate (Kitaw et al. 2007). This, the report stated, had resulted in increased workloads for the available trainers and caused difficulties in supervision during practical placements and internships (Kitaw et al. 2007). A general shortage of trainers also impacted on the EHRP in Malawi (Management Sciences for Health 2010). Different strategies were adopted to address these obstacles. Although experienced health in Malawi were seconded to join as trainers (Management Sciences for Health 2010), the significant salary disparity between practitioners and trainers at these institutions proved to be an important disincentive (Paul Marsden 2012, personal communication). In Zimbabwe, a cascade training model was used where tutors trained other trainers from different provinces. This is reported to have facilitated the rapid scale-up of training (David Sanders 2012, personal communication). The quality and appropriateness of training are affected by the training conditions and content. In addition to the impacts caused by shortages of trainers, the quality and John Wiley & Sons Ltd

9 content of the training also negatively affected several programmes. In Ethiopia, most lay health worker trainers had only a limited knowledge of the local languages. The use of foreign trainers and English as the medium of instruction were barriers to the effective delivery of training (Kitaw et al. 2007). Success was also limited by factors including a lack of materials, such as textbooks and demonstration materials, and inadequate infrastructure, such as classrooms, latrines and water supplies (Kitaw et al. 2007). In India, ASHAs in several states reported that they felt uncomfortable with the language of instruction, and that the sessions were too crowded and therefore not conducive to learning (Bajpai & Dholakia 2011). In Iran, the training of behvarzes is delivered by specialised training centres (District Behvarz Training Centre) managed by the district health system (Javanparast et al. 2012). A comprehensive curriculum with relevant topics, highly qualified trainers, good trainer trainee relationships and clinical placements under direct supervision of trainers were factors that were seen to have ensured that the pre-service training of behvarzes was of high quality (Javanparast et al. 2012). However, with regard to in-service training, the same study reported that behvarzes complained about several factors, such as poor quality and timing, courses running infrequently and inadequately qualified trainers who were unfamiliar with the working conditions of behvarzes (Javanparast et al. 2012). The study suggested that greater involvement of the specialised training centres in delivering in-service training could improve both quality and satisfaction among behvarz. In Venezuela, difficulties in teaching curricula in a way that were easily understandable to lay health, who often only had primary education or did not speak Spanish as their first language, were reported from the Medicina Simplificada Program (Yates 1975). The training provided for health has also not always been appropriate to the tasks they are expected to undertake. Two evaluations from 2008 reported that lay health in Ethiopia (known locally as Health Extension Workers) felt insufficiently trained to assist women during delivery even though this task was part of their job (Banteyerga & Kidanu 2008; Federal Ministry of Health 2008). A more recent study from Ethiopia (Medhanyie et al. 2012) reported that levels of knowledge of antenatal care and pregnancy among Health Extension Workers have remained poor (Medhanyie et al. 2012) and that this may still partially be due to inadequate levels of training. Another evaluation from 2011 (Center for National Health Development in Ethiopia 2011) reported that the majority of Health Extension Workers surveyed believed that the responsibilities entrusted to them required additional training. The study found that the utilisation of delivery and post-partum care at the health posts was low and suggested that this may have been due to a lack of refresher training and inadequate skills among the Health Extension Workers (Center for National Health Development in Ethiopia 2011). In India, an evaluation completed in 2011 reported that ASHA training modules covering the topics of reproductive, child health and nutrition were incomplete and did not adequately match the tasks assigned to them within their community roles (National Health Systems Resource Centre 2011). To address these training gaps, some Indian states strengthened the course content by supplementing it with their own additional modules and including other topics (National Health Systems Resource Centre 2011). Studies of the Primary Health Care in Iran reported a different strategy for ensuring the appropriateness of training: training policies and content were updated regularly through national and provincial meetings, in response to both changing health needs and increasing levels of literacy among the training candidates (Javanparast et al. 2011a, 2012). An attempt was made in Brazil in 2004 to ensure consistent quality in the training of lay health (a cadre known locally as Community Health Agents) through the development of a national reference guide (Camila Giugliani 2012, personal communication). However, because the implementation of the training courses was a local responsibility, the success of this initiative has been varied (Camila Giugliani 2012, personal communication). Management and programme support structures Most of the examined programmes had attempted to develop or use existing local management structures, such as district health offices. Typically, these management structures have been tasked with monitoring programmes at a village and district level and with ensuring that particular support functions are carried out (Table 1). However, according to the evaluations, we identified a number of management problems had affected programme implementation, and these are described below. Levels of staffing and management skills may affect programme functioning. District health offices in Ethiopia (known locally as Woreda Health Offices) were given the responsibility of coordinating the Health Services Extension and helping Health Extension Workers to implement programme activities (Federal Ministry of Health 2005). Several studies reported, however, that the Woreda Health Offices lacked sufficient training capacity due to factors such as understaffing 2014 John Wiley & Sons Ltd 1445

10 (Teklehaimanot et al. 2007; Banteyerga & Kidanu 2008). This had resulted in irregular support and supervision for Health Extension Workers and impacted upon the monitoring and coordination of the programme (Teklehaimanot et al. 2007; Banteyerga & Kidanu 2008). A 2010 evaluation of the EHRP in Malawi reported that many managers had been assigned clinical responsibilities in addition to their standard duties. Many also felt ill-prepared when managing the core human resource functions stipulated by the Ministry of Health. This was because the managers faced challenges such as understaffing, a lack of job satisfaction, low morale, poor working conditions, and a lack of adequate skills and training (Management Sciences for Health 2010). Understaffing and high vacancy rates in management structures had also led to underqualified and inexperienced staff being employed as stand-ins for public sector positions (Paul Marsden, personal communication). Malawi s 2006 human resource strategy document proposed that a scale-up of the number of health should be matched by a corresponding scale-up in management and leadership capacity. In reality, this was difficult to achieve due to other funding demands and priorities (Paul Marsden 2012, personal communication). A 2011 evaluation of the National Rural Health Mission in India reported that a lack of investment by states in different levels of the management structure had resulted in weak support systems for ASHAs (ASHA 2011). An inadequate support system and poor supervision of lay health were also reported in the Primary Health Care in Iran (Javanparast et al. 2011b). In Brazil, municipalities are responsible for most management and implementation decisions concerning the Family Health (Magalh~aes & Senna 2006; Rocha & Soares 2009), aided by the support of state- and national-level health departments (Dr Camila Giugliani 2012, personal communication). However, many municipal health managers have lacked the necessary qualifications (Dr Camila Giugliani 2012, personal communication). Problems with the management and disbursement of funds at the local level may negatively affect programme implementation. How funds are managed and disbursed emerged as an important factor affecting implementation. India s National Rural Health Mission, for instance, allocates flexible financing ( untied funds ) to subcentres, primary health centres and community health centres for the maintenance of healthcare facilities (National Rural Health Mission 2005a; Bajpai et al. 2009; Gill 2009). How these grants are utilised should be determined through regular meetings held by Village Health Committees and hospital societies (National Rural Health Mission 2005a; Bajpai et al. 2009). Evaluations from 2009 and 2011 indicated that members of the Village Health Committees, medical officers, hospital supervisors and senior administrators were not always certain about the mechanisms for spending the funds nor what to spend them on. Formal management guidelines were often lacking (ASHA 2011; Bajpai et al. 2009; Evaluation Organisation 2011) and several evaluations reported poor use of funds at district level (Bajpai et al. 2009; Gill 2009; National Rural Health Mission 2010). Payment of health was also noted as a problematic management issue. Several evaluations of the National Rural Health Mission described delays in sending incentive, stipend and salary payments to ASHAs (ASHA 2011; Bajpai & Dholakia 2011; Bajpai et al. 2009; National Rural Health Mission 2010; Evaluation Organisation 2011). One evaluation identified a number of procedural issues affecting payment, such as funds not being transferred to subdistricts, lack of familiarity of managers with e-banking and confusion about which incentives were available (Bajpai & Dholakia 2011). The organisation and delivery of services Referral systems, the relationships between different cadres, and the number of lay health deployed in communities were factors found to affect the organisation and delivery of health services. The distribution density of lay health sometimes affects their ability to deliver health care. Several programmes reported that a low distribution density of lay health could lead to work overload and the neglect of specific tasks, particularly in areas in which households are widely scattered or health have to cover large distances. In Zimbabwe, the intended ratio of Village Health Worker to villagers was 1:500 to 1000 people. Providing the necessary training to achieve this ratio was difficult. In some areas, health were responsible for six villages and approximately 6000 people (Sanders 1992). Where Village Health Workers were widely dispersed, they typically had to cover greater distances to reach communities within their catchment areas. Data suggested that more attention would probably have been given to maternal and child health if each Village Health Worker had worked in a geographically bound community (Sanders 1992). In Ethiopia, Health Extension Workers were similarly assigned a number of John Wiley & Sons Ltd

11 villages and scattered subvillages: cadres had to walk 3 4 h between the subvillages and this meant that they had less time available for work and were faced with increased work pressures (Banteyerga & Kidanu 2008; Federal Ministry of Health 2008). Larger catchment area sizes were also found to affect community outreach for lay health worker programmes in India (ASHA 2011) and Brazil (Fausto et al. 2011). The quality of relationships between different cadres varies and this impacts upon health worker performance. The social dynamics between different cadres can influence task-shifting initiatives (Callaghan et al. 2010; Reeves et al. 2010), and a number of factors were found to influence the level of cooperation between healthcare providers. Clinical officers in Malawi expressed frustration over their salary levels, benefits, workload and status compared to those given to doctors (Bradley & McAuliffe 2009). A study from Brazil reported difficulties in integrating lay health into family health teams because of poor communication and low levels of cooperation between lay health on one hand, and physicians and nurses on the other (Zanchetta et al. 2009). Likewise, it was reported that the Medicina Simplificada Program in Venezuela had difficulties initially in convincing rural physicians to work closely with the lay health linked to the programme (Yates 1975). A study from India suggested that meaningful communication between ASHAs and other cadres was impeded by the professional hierarchies which limited interactions between ASHAs and other more specialised cadres (Scott & Shanker 2010). Another evaluation from India reported confusion over the responsibilities designated to Anganwadi (a type of local lay health worker) and ASHAs (Bajpai & Dholakia 2011) and a lack of clarity regarding their respective roles. Inadequate referral systems negatively impact on programmes. A well-functioning referral system is seen as a key element of most task-shifting initiatives (Zachariah et al. 2009; Kane et al. 2010), but data related to this issue is limited and, in this review, came from only two country settings. Evaluations of the Health Services Extension in Ethiopia reported that the liaison between and their nearest health facilities was poor. Health Extension Workers rarely received feedback about the referrals they sent to the health posts (Teklehaimanot et al. 2007; Banteyerga & Kidanu 2008; Bekele et al. 2008; Center for National Health Development in Ethiopia 2011). Furthermore, referral processes were found to be constrained by large distances, poor roads and by the scarcity or high cost of transport (Teklehaimanot et al. 2007; Banteyerga & Kidanu 2008; Center for National Health Development in Ethiopia 2011). A lack of health centres as well as shortages of clinical nurses, health officers, medical supplies and equipment were also seen as important limiting factors (Teklehaimanot et al. 2007; Banteyerga & Kidanu 2008; Center for National Health Development in Ethiopia 2011). In India, programme guidelines for the National Rural Health Mission specify that transport services for institutional delivery are to be provided as part of wider improvements to the referral system (National Rural Health Mission 2005a,b). However, evaluations of the National Rural Health Mission have noted that the availability of rural emergency transport varies between states (National Rural Health Mission 2010; Evaluation Organisation 2011), and have suggested that the facility-based services for managing referrals from ASHAs are still in need of improvement (ASHA 2011). Community participation and sociopolitical factors Many of the programmes sought to decentralise decisionmaking and ensure better community participation and ownership through local citizen organisations. The responsibilities of these bodies included undertaking programme planning and programme activities. Several challenges to ensuring meaningful community participation during the implementation of the programmes were identified. The establishment of village and district committees is often slow or delayed. In Ethiopia, community-level programme planning is meant to be guided by Village Health Committees, and membership includes Health Extension Workers and community representatives (Federal Ministry of Health 2005). One study noted that only 25% of the villages sampled from six regions had village committees established (Teklehaimanot et al. 2007). Another study, conducted in four of these regions a year later, reported that although some improvements had been made, further strengthening of the village structures was needed (Banteyerga & Kidanu 2008). Village Health and Sanitation Committees are local citizen bodies responsible for supporting ASHAs at the community level in India (National Rural Health Mission 2005a). Members include local government representatives, local health (including ASHAs and community representatives) and health volunteers (National Rural Health Mission 2005a). Evaluations have indicated that the process of establishing the committees in some states had been slow (Bajpai et al. 2009; Evaluation Organisation 2011), which had negatively 2014 John Wiley & Sons Ltd 1447

12 affected the management of the programme. More recent work has suggested that the committees are now established in most villages ( Evaluation Organisation 2011). Existing village and district committees sometimes function poorly. Local committees may not always function as intended and health and community members may be uninvolved or unaware of them. Reports from India suggest that in many villages, meetings of the Village Health and Sanitation Committees are not held regularly and that the support they give to ASHAs and to the use of local funds for equipment purchases has varied (Srivastava 2008; Bajpai et al. 2009; Evaluation Organisation 2011). An evaluation in Ethiopia conducted in 2008 also observed that lay health were not always represented in village management structures (Teklehaimanot et al. 2007). This problem, however, appears to have improved subsequently and most Health Extension Workers were in 2011 reported to be members of the Village Committees (Center for National Health Development in Ethiopia 2011). A 1983 evaluation of Zimbabwe s Village Health Worker found that fewer than 30% of the Village Health Workers (n = 95) were aware of the existence of the Health Committees in their local areas (Sanders 1992), thus suggesting that some of the committees were not functioning properly. Links between communities and the Village Health Workers also appeared to be weak: only a third of the community members reported being involved in the selection of health (Sanders 1992). Similar weaknesses were reported in Iran. Here, as part of the country s primary healthcare programme, community councils (known locally as a behvarz councils) had been established to engage behvarzes in service planning, knowledge transfer and the development of good practice (Javanparast et al. 2011b). One study noted that most behvarzes believed that the aims of these councils had not been adequately realised (Javanparast et al. 2011b). Establishing meaningful community involvement also appeared to be difficult in Brazil. There, local health councils have been established as a forum for community members and community health agents. These councils aim to facilitate community participation in defining priorities for the Family Health at the local level and in planning service delivery (Magalh~aes & Senna 2006; Peres et al. 2006). One study of six municipalities (Peres et al. 2006) indicated that many Family Health Team members never participated in their local health councils, and that on average, Family Health Team members participated in meetings with community members 4 days per year (Peres et al. 2006). This level of contact was seen as inadequate for ensuring that the Teams were integrated closely within the community. Lack of participation in these forums may impede community health agents from fulfilling their role as an agent for identifying local health needs, and mobilising and facilitating social participation and intersectoral action (Fausto et al. 2011). The extent to which government programmes are locally owned is debatable. The purpose of organisational structures such as the Village Committees in India is to ensure greater community participation and ownership in healthcare delivery. However, community members have not always regarded these structures as inclusive. For instance, a 1982 evaluation from Zimbabwe found that almost half of community members interviewed (n = 121) had not heard of the National Village Health Worker (Sanders 1992), and a similar lack of programme awareness was reported by the National Rural Health Mission in India ( Evaluation Organisation 2011). The Zimbabwean study included in this synthesis explored the connection between the payment of Village Health Workers and levels of community accountability, noting that if lay health were to remain responsive to the communities they worked in, then the payment they received should ideally be administered by the communities themselves. The lack of resources in many communities resulted in the Zimbabwean central government providing funding for the National Village Health Worker via District Councils (Sanders 1992). Community members therefore came to regard the contribution of the Village Health Workers as an additional component of the public services the communities were receiving. As such, the Village Health Workers were perceived as being accountable largely to the government rather than to the actual communities (Sanders 1992). An early evaluation of the Medicina Simplificada Program in Venezuela, also suggested that the combination of problems caused by economic disadvantage in communities and by centralised governance appeared to have impeded community involvement (Gonzales 1975). Community awareness of the Ethiopian programme varied between regions but, overall, the evaluations suggested that awareness about the programme was increasing (Banteyerga & Kidanu 2008; Center for National Health Development in Ethiopia 2011). Evaluations from Zimbabwe further suggested that existing community initiatives created through social and popular mobilisation may be adversely affected by largescale programmes (Sanders 1992). Prior to the scale-up John Wiley & Sons Ltd

13 Table 3 Where health systems need to be strengthened Linking the findings of the country case synthesis to the WHO health systems building blocks* *The WHO health systems building blocks provide a framework for understanding the necessary components of a functional health system (73). Issues in italics cut across more than one WHO health systems building block John Wiley & Sons Ltd 1449

14 Table 4 Comparison of the synthesis findings with those of other related reviews Review Bhutta et al. (2010) Celletti et al. (2010) Dawson et al. (2013) Fulton et al. (2011) Jaskiewicz and Tulenko (2012) Review aims, methods and included cadres Aim: To identify lay health worker programmes which had a positive impact on achieving the Millennium Development Goals related to health, and to explore country experiences with lay health Methods: Country case studies of programmes in eight LMICs, using published and unpublished reports and direct contact with key personnel Aim: To identify the critical elements of an enabling environment which can ensure that lay health provide quality HIV services in a manner that is sustainable Methods: Desk review of country-specific documents on lay health programmes, review of national laws and regulations, and interviews with key informants Aim: To determine evidence to optimise health worker roles through task shifting/sharing to address Millennium Development Goal 5, reduce maternal mortality and provide universal access to reproductive health Methods: Narrative synthesis of peer-reviewed literature ( ) on task-shifting activities by doctors, doctor assistants, nurses, midwives, auxiliary nurses, auxiliary midwives and lay health. The outcomes assessed included heath worker performance, patient outcomes, provider needs and experiences, and cost-effectiveness. The review included observational, quasi-experimental and descriptive studies Aim: To review the health workforce skill mix literature to determine the strength of the evidence, identify gaps in the evidence, and to propose a research agenda Methods: Literature review using an economic perspective. Included a range of study designs (RCTs, quasi-rcts, case studies, qualitative studies and literature reviews). The analysis process was not described clearly Aim: To examine the working conditions of lay health and the impacts these may have on work productivity Method: Review of selective published and unpublished articles and reports on lay health worker programmes in low- and middle-income countries to explore factors influencing lay health worker productivity A comparison of the review findings with the findings of this synthesis Only considered lay health Included two of the programmes in this review (Brazil, Ethiopia) Main output of the country case studies was the documentation of how the lay health worker programmes were organised and the extent to which these programmes were functional (assessed using a functionality tool), rather than an exploration of the factors affecting implementation at scale Included two of the programmes in this review (Brazil, Ethiopia) Focused on how lay health can provide quality HIV services Some themes, such as training and supervision, overlapped with this review while other themes, such as management issues and community participation, were not addressed Did not take a programme-based case study approach to task shifting at scale, but looked at data from evaluations and descriptive studies in the peerreviewed literature Some of the barriers to task shifting that were identified overlapped with the findings of this synthesis. These included: poor skills, inadequate staff training and coordination, and a lack of equipment and supplies in some settings. System-level factors identified in this synthesis, such as those related to financing, governance and community engagement, were not discussed extensively in the review Focused on low-income countries only rather than LMICs Did not take a programme-based case study approach, but looked instead at data from multiple single studies and reviews Findings shared some themes with this synthesis with regard to the impact of context on the success of task-shifting initiatives, and concerns regarding quality, safety and sustaining performance and motivation. Focused more on cost and costeffectiveness issues than on systems issues Only considered factors affecting the productivity of lay health and therefore provides more depth about lay health worker issues than this synthesis Did not include data on other cadres Findings share themes with this synthesis but did not generally explore system-level factors, such as the financing and governance of programmes at scale John Wiley & Sons Ltd