Development assistance for community health workers in 114 low- and middle-income countries, 2007–2017

Abstract Objective To estimate the level and trend of development assistance for community health worker-related projects in low- and middle-income countries between 2007 and 2017. Methods We extracted data from the Organisation for Economic Co-operation and Development’s creditor reporting system on aid funding for projects to support community health workers (CHWs) in 114 countries over 2007–2017. We produced estimates for projects specifically described by relevant keywords and for projects which could include components on CHWs. We analysed the pattern of development assistance by purpose, donors, recipient regions and countries, and trends over time. Findings Between 2007 and 2017, total development assistance targeting CHW projects was around United States dollars (US$) 5 298.02 million, accounting for 2.5% of the US$ 209 277.99 million total development assistance for health. The top three donors (Global Fund to Fight AIDS, Tuberculosis and Malaria, the government of Canada and the government of the United States of America) provided a total of US$ 4 350.08 million (82.1%) of development assistance for these projects. Sub-Saharan Africa received a total US$ 3 717.93 million, the largest per capita assistance over 11 years (US$ 0.39; total population: 9 426.25 million). Development assistance to projects that focused on infectious diseases and child and maternal health received most funds during the study period. Conclusion The share of development assistance invested in the CHW projects was small, unstable and decreasing in recent years. More research is needed on tracking government investments in CHW-related projects and assessing the impact of investments on programme effectiveness.

One of the most important challenges for successfully scaling-up CHW programmes is obtaining sufficient and sustainable funding to support the programme. [3][4][5] A study has estimated that deploying CHWs in rural areas of sub-Saharan African countries requires at least 2.6 billion United States dollars (US$), funded either by national governments or donor partners or both. 20 Other analyses suggest even higher figures. 5,21 Knowing how much has been invested in CHW programmes is important for policy-makers, donors and other stakeholders to make evidence-based planning and budgeting. Our literature review for this study demonstrates that the empirical evidence about existing investments is scant at best. We aimed to assess the level and trend of development assistance for CHW programmes. International donor contributions play an important role in financing the health sector of low-income countries. 22 Estimating development assistance for CHW programmes is necessary for establishing a financial baseline for such programmes and identifying funding gaps. Knowing who are the top investors in CHW programmes will help locate potential sources of funds for scaling-up these programmes in the future. Here, we provide a descriptive analysis on the flows of total and per capita development assistance for CHW programmes by donors, recipient regions and recipient countries. We also investigated the priorities of the development assistance for CHWs and estimated per capita aid for CHW disbursed to low-and middle-income countries between 2007 and 2017.

Data sources
We extracted data from the aid activity database from the creditor reporting system of the Organisation for Economic Co-operation and Development's development assistance committee. 23 The creditor reporting system database is publicly accessible and has been widely used in aid tracking studies. [24][25][26][27][28][29][30] The data are reported directly by the governments of the 30 members of the development assistance committee, 65 multilateral organizations (e.g. the World Bank group) or global Research Development assistance for community health workers Chunling Lu et al.
health initiatives (e.g. the Global Fund to Fight AIDS, Tuberculosis and Malaria), 25 non-development assistance committee countries (e.g. United Arab Emirates) and 28 private donors (e.g. Bill & Melinda Gates Foundation). We downloaded the data used in this study in July 2019. We included 119 countries classified as low-and middle-income by the World Bank in 2015 31 (available in the data repository). 32 These were countries that had received aid for CHW projects from 41 donors since 2007 and had available annual data on population and on World Bank region classifications. 33,34 Definitions We followed the World Health Organization (WHO) definition of CHWs as: health workers who have been trained to some extent, but do not possess a formal professional certificate. 15 Examples include village health workers, peer supporters, community volunteers and health extension workers. Meanwhile, we were aware of some programmes that aimed for more formalization, accreditation or certification of this role. 35,36 We defined development assistance as aid disbursed to projects that support CHW programmes in low-and middle-income countries. In the creditor reporting system data, the term projects refers to programmes and activities that are supported with development assistance.
There were no indicators in the creditor reporting system data on projects for CHWs. Following previous practice, [25][26][27][28][29][30] we constructed a list of keywords (Box 1), based on a review of keywords used in other literature to identify projects for CHW programmes. Our focus was on projects in three health-related sectors provided by the creditor reporting system data: (i) general health; (ii) basic health; and (iii) population policies or programmes and reproductive health.
As some projects on communitybased interventions or community health might not have included CHWrelated keywords in their project descriptions, we generated two sets of data on development assistance for CHW programmes. To obtain the lower bound of estimates, we calculated aid that targeted CHW projects, as specifically defined by the keywords in community health worker-targeted projects listed in Box 1. To obtain the upper bound of estimates, we calculated aid that included both CHW-targeted projects and community intervention projects that were not defined by these keywords, but which included keywords indicating CHW components, as described in the Community health worker-inclusive projects listed in Box 1.
Some projects may have included other components that were not related to CHWs. In this case, we were not able to allocate funds to the CHWs component due to lack of information in the data. We therefore considered aid to those projects as investments in CHW projects. Details of the keyword search and our validation strategy are available in the data repository. 32

Analysis
We used actual disbursements to estimate development assistance for CHWs between 2007 and 2017. This time frame allowed us to avoid the issue of missing disbursement data in the creditor reporting system: the missing rate of total disbursements in three health sectors was approximately 23% on average between 2000-2006. 28 The completeness of disbursement data has remained at almost 100% since 2007. 37 A total of 81.0% (US$ 196 539.00 million) of US$ 242 598.30 million development assistance for health during this period was in the form of grants and 5.7% (US$ 13 897.26 million) as concessional loans. We report all disbursements in 2017 US$.
We estimated levels and trends of development assistance for CHWs over 2007-2017 in both total and per capita spending at the global, regional and country levels. We identified the top 10 donors of development assistance for the CHW projects and the 10 CHW-targeted projects that had received the largest disbursements. When we calculated country-level estimates, we excluded projects that were classified as going to a region only since no further information Box 1. Keywords used to search for community health worker-related projects in the creditor reporting system Research Development assistance for community health workers Chunling Lu et al.
was provided on the amounts provided to each country in that region.
In the three health sectors that we included in our analysis (general health; basic health; and population policies or programmes and reproductive health), projects were categorized by the creditor reporting system into 17 primary purposes, such as sexually transmitted diseases including human immunodeficiency virus and acquired immune deficiency syndrome (HIV/AIDS), basic nutrition, and so on. 37 Further details are available in the data repository. 32 We excluded 98 projects whose primary purpose was research, such as randomized trials. After excluding one purpose (medical research), we finally assessed the allocation of development assistance for CHW-targeted projects across the 16 purposes, with 4600 CHW-targeted projects and 10 363 CHW-inclusive projects. We analysed development assistance for projects that aimed to improve the health-related millennium development goals (MDGs) 38 and grouped them into two areas: (i) HIV, tuberculosis and malaria; and (ii) child and maternal health, including projects on family planning, reproductive health, basic nutrition, and child, maternal and newborn health. We used variables on the primary purposes of a project and adopted keywords from previous studies to further identify the projects in these two areas (available in the data repository). 32 Note that development assistance for CHW projects in these two areas are not mutually exclusive; a project targeting infectious diseases may also include support for child and maternal health.
In this article we present the development assistance for CHW-targeted projects in 114 countries. The results of development assistance for CHWinclusive projects in 119 countries are available in the data repository. 32

Results
Between 2007 and 2017, an estimated total of US$ 5298.02 million was disbursed to 4600 CHW-targeted projects.

Estimates by donor
Among the 37 donors that disbursed funds to CHW-targeted projects, the top 10 and their annual contribution are presented in Fig. 1. The top three donors during the study period were the Global Fund (US$ 2 718.09 million), government of Canada (US$ 900.39 million), and government of the United States of America (US$ 731.60 million). Altogether they provided 82.1% of total development assistance for CHWtargeted projects.

Estimates by region
Among the six regions analysed (World Bank classifications), sub-Saharan African countries consistently received the largest amount of development assistance (US$ 3 717.93 million in total over the 11 years), accounting for 70.2% of total development assistance disbursed to CHW-targeted projects (Fig. 2). When taking population size into consideration (estimated total population over 11 years: 9 426.25 million), sub-Saharan African countries had the largest per capita development assistance (US$ 0.39) during the study period.

Estimates by purpose
Among the 16 health purposes labelled by the creditor reporting system data, the projects with a focus on three infectious diseases (HIV, tuberculosis and malaria) and child and maternal health received the most funds. For example, the category sexually-transmitted disease control including HIV/AIDS received the largest amount (US$ 2 061.14 million) and percentage of total disbursements (38.9%) during the study period, followed by malaria control (US$ 1 047.46 million; 19.8 %) and reproductive health care (US$ 490.85 million; 9.3%; Table 2).
The 10 projects that received the largest disbursements during the period were in sub-Saharan African countries. Eight of them were funded by the Global Fund. The other two donors were the International Development Association of the World Bank and the government of the United States. Seven projects had a focus on HIV, two on malaria and one on child and maternal health (Table 3).

Estimates by country
Among the 25 countries that received more than US$ 0.50 per capita aid for CHW over the 11 years, 12 were low-

Discussion
Our analysis of data from the creditor reporting system on aid disbursed to projects that supported CHWs between 2007 and 2017 revealed two main findings. First, development assistance for these projects accounted for less than 3% of total development assistance for health during the period, with a decrease since 2013 mainly due to the reduction of Global Fund investments in these projects. Second, the focus of CHW-targeted projects was mostly on achieving the health-related MDGs (HIV, tuberculosis, malaria and child and maternal health) and most of the assistance was disbursed to sub-Saharan African countries. This finding suggests that donors' investment in CHWs was in line with efforts towards achieving the health-related MDGs in the region. There is increasing evidence of the effectiveness of CHWs in contributing to improved health outcomes for children and women. For example, a review of efficacy studies indicated that CHWs could reduce maternal mortality by 42-78%. 39 A review of studies on the ef-ficacy of CHWs' prescribing antibiotics for treatment of pneumonia (one of the leading causes of mortality among infant and children younger than 5 years) indicated that under-5 mortality could be reduced by 13-60%. 40 The long-term impacts of CHWs on social and economic development, including empowering women and increasing jobs especially in remote poor and rural areas, could result in a positive economic return, as high as 10:1. 3 There is now a broad global consensus that CHWs are essential for efforts to achieve the 2030 health-related sustainable development goal (SDG) targets, including the achievement of universal health coverage. 7,8,[41][42][43][44] Our study shows that the average annual amount of development assistance for CHW-targeted projects in sub-Saharan African countries over 2007-2017 was US$ 0.34 billion. It has been estimated that the governments of sub-Saharan African countries spent US$ 0.4 billion of their own funds on CHWs. 21 The sum of these two estimates falls far short of the estimates required for CHWs to be minimally effective (US$ 2.6-3.1 billion 5,20,21 ), highlighting an urgent need to build national political and financial support for CHWs in the region.
In the short term, however, lowincome countries are not able to fully support a national CHW programme with domestic public spending alone. To illustrate the importance of donor investment in low-income countries we made a case study of Rwanda. Box 2 shows that the country's CHW programmes, which played an essential role in provide basic care for child and maternal health, were mainly funded by foreign aid. The programme made important contribution to reaching the country's health MDGs, 3 even with less than half the recommended level of per capita investment (US$ 3.25 versus US$ 6.86). 20 In our study, among the 25 countries that received the largest per capita aid for CHWs over the 11 years, only 12 were low-income countries.
A marked fluctuation of the development assistance for CHW-targeted projects over time was mainly driven by changes in investments from one or two donors. Relying on a few donors poses challenges for maintaining stable funding levels and effective long-term budgeting for recipient countries. Development assistance for community health workers Chunling Lu et al.
Our study had limitations. First, the creditor reporting system data do not include aid from many private nongovernmental organizations as well as from emerging economies such as China. To address this issue, we will conduct future searches on the websites of nongovernmental organizations that are known for supporting CHWs in recent years (e.g. the Clinton Health Access Initiative and Global Health Workforce Alliance). Second, our identification strategy relied on keywords, which were not able to capture all projects on development assistance for CHW programmes. However, we estimate that approximately 5% of projects were missed, indicating that our findings are robust (available in the data repository). 32 Third, the accuracy of our data could suffer from errors or incompleteness in the reporting of the characteristics of the projects supported through development assistance for health. Fourth, for projects with multiple foci, we were unable to determine how much money went to each focus, so we considered that the total funds went to support CHW programmes. We may therefore have overestimated the amounts donated for CHW programmes. Fifth, when estimating development assistance for CHWs at the country level, project funds that were disbursed to an entire region were not included because of a lack of information about how to allocate funds to each country.
Several lessons can be drawn from our study. First, our finding that less than 3% of total development assistance for health was targeted on CHW programmes indicates that there is an opportunity to improve the effectiveness of this assistance by investing more in programme expansion and strengthening. This finding is important as there is extensive evidence about the effectiveness of CHWs in improving maternal, neonatal and child health, including control of malaria and expanding contraceptive access. [48][49][50][51] There is a continuing need to reduce the number of readily preventable deaths in these population groups to achieve the global goal of eliminating preventable maternal and child mortality by 2030. 52,53 Second, it is concerning that development assistance for CHWs has been decreasing since 2014 and only accounted for 1.27% of total development assistance for health in 2017. The reduction was due to a decrease in support from the three main donors of the development assistance for CHWs. There needs to be more stability of CHW programme assistance over time, support from a broader set of donors and a broader focus of the support beyond high-priority infectious diseases in the era of SDGs.
Third, further research is needed into the amount of investment going into CHW programmes -both from external development assistance as well as from governments' own internally generated funds -and the proportion of investment going for CHW programmes relative to other areas of investment in health. Investments in CHW projects is an important but neglected area of research.
Finally, we need evidence of the effectiveness of development assistance for CHWs in terms of both the role of aid for these programmes in improving population health outcomes and the best ways to support the programmes. The future research would include monitoring investments in CHW programmes and assessing their impact on CHW programme effectiveness. This information will help policy-makers and other stakeholders plan for long-term financing strategies for scaling-up and sustaining CHW programmes.

Box 2. Case study: financing community health worker programmes in Rwanda
Established in 1995, Rwanda's CHWs have become an essential part of the Rwandan health system and have played an important role in expanding service coverage in Rwanda, especially in remote rural areas. 3 According to a government report in 2016, about 45 000 CHWs served 11 million people. 45 The CHWs were providing services reliably in each village, including health promotion (e.g. healthy living), disease prevention (e.g. bed-net utilization) and treatment (e.g. treatment of malaria). Major activities included promoting child and maternal care (e.g. antenatal home visits and treating or referring young children with diarrhoea or malnutrition) and treating tuberculosis and malaria. Health resource tracking data in Rwanda provides project-level expenditure information by funding sources. 46 We used data that were available in fiscal years 2010-2011 and 2011-2012 to estimate the spending on CHWs by the Rwanda government and foreign donors in the two years. Funds were calculated in 2010 US$.

Aide au développement pour les agents de santé communautaires dans 114 pays à revenu faible et intermédiaire, de 2007 à 2017 Asistencia para el desarrollo de los trabajadores de salud de la comunidad en 114 países de ingresos bajos y medianos, 2007-2017
Objetivo Estimar el nivel y la tendencia de la asistencia para el desarrollo destinada a proyectos relacionados con los trabajadores de salud de la comunidad en países de bajos y medianos ingresos entre 2007 y 2017. Métodos Se extrajeron datos del sistema de información de acreedores de la Organización para la Cooperación y el Desarrollo Económicos sobre la financiación de la ayuda para proyectos de apoyo a los trabajadores de salud de la comunidad (community health workers, CHW) en 114 países entre 2007 y 2017. Se elaboraron estimaciones para los proyectos descritos específicamente mediante palabras clave pertinentes y para los proyectos que podrían incluir componentes sobre los CHW. Se analizó la pauta de la asistencia para el desarrollo por finalidad, donantes, regiones y países receptores, y las tendencias a lo largo del tiempo. Resultados Entre 2007 y 2017, el total de la asistencia para el desarrollo destinada a proyectos de CHW fue de unos 5 298,02 millones de dólares estadounidenses (USD), lo que representa el 2,5 % del total de la asistencia para el desarrollo destinada a la salud, que ascendió a Research Development assistance for community health workers Chunling Lu et al.