Positive spill-over effects of ART scale up on wider health systems development: evidence from Ethiopia and Malawi

Background Global health initiatives have enabled the scale up of antiretroviral treatment (ART) over recent years. The impact of HIV-specific funds and programmes on non-HIV-related health services and health systems in genera has been debated extensively. Drawing on evidence from Malawi and Ethiopia, this article analyses the effects of ART scale-up interventions on human resources policies, service delivery and general health outcomes, and explores how synergies can be maximized. Methods Data from Malawi and Ethiopia were compiled between 2004 and 2009 and between 2005 and 2009, respectively. We developed a conceptual health systems framework for the analysis. We used the major changes in human resources policies as an entry point to explore the wider health systems changes. Results In both countries, the need for an HIV response triggered an overhaul of human resources policies. As a result, the health workforce at health facility and community level was reinforced. The impact of this human resources trend was felt beyond the scale up of ART services; it also contributed to an overall increase in functional health facilities providing curative, mother and child health, and ART services. In addition to a significant increase in ART coverage, we observed a remarkable rise in user rates of non-HIV health services and an improvement in overall health outcomes. Conclusions Interventions aimed at the expansion of ART services and improvement of long-term retention of patients in ART care can have positive spill-over effects on the health system. The responses of Malawi and Ethiopia to their human resources crises was exceptional in many respects, and some of the lessons learnt can be useful in other contexts. The case studies show the feasibility of obtaining improved health outcomes beyond HIV through scaled-up ART interventions when these are part of a long-term, system-wide health plan supported by all decision makers and funders.

of service delivery platforms if patients with HIV and other chronic illnesses are to be appropriately managed [2].
Th e impact of new additional fi nancing for HIV programmes on health systems and non-HIV-related health services has been discussed extensively in recent years. Some assert that new additional fi nancing, while enabling the scale up of HIV services, has overburdened already weak health systems, and thus worsened access for the target population [3,4]. In contrast, others indicate that HIV programmes and funding may also have a positive impact on health systems by reinforcing health support functions, such as human resources, infrastructure, laboratories and drug supply [5,6,7]. Yet, few studies have systematically explored the possible positive spill-over eff ects of HIV/AIDS programmes on health systems [8].
Th is study analyzes the eff ect of ART scale-up interventions on: (i) human resources; (ii) non-HIV service delivery platforms; and (iii) non-HIV health outcomes and goals, mainly focusing on primary health and maternal and child care indicators. Th e study draws on evidence from Malawi and Ethiopia: these countries, both confronted with a high number of people in need of ART, have successfully scaled up ART services in recent years, in spite of initially weak health systems. Th e study focuses on the impact of HIV programmes and funds on human resources for health and subsequently on health outcomes, using an impact pathway derived from a health systems conceptual framework by van Olmen et al [9].
Malawi and Ethiopia are low-income countries with, respectively, populations of 12.6 million and 73 million, and estimated HIV prevalence of 14.4% and 4.6% (Table 1); each country had more than a million people living with HIV in 2004 at the start of the HIV programme expansion. In spite of domestic and international eff orts, health systems in these countries remain weak and underfunded. Hence, there are chronic shortages of human resources, infrastructure and essential drugs. Malawi and Ethiopia have some of the lowest ratios of medical doctors per 100,000 people in the world as a result of low training output, a relatively high brain drain and the severe adverse impact of HIV on the health workforce [13].
To scale up ART, both countries adopted a communityfocused public health model. ART care activities were gradually decentralized to peripheral health centres and integrated into primary healthcare, which implied simplifi cation, standardization and rationalization of treatment protocols. Th is was accompanied by a shift of several clinical and administrative tasks to lower-level health cadres and the gradual involvement of the community in ART care delivery [14][15][16].
In 2004, the HIV epidemic in Malawi highlighted the crisis of human resources for health, and triggered the introduction of an "Emergency Human Resources Program". Th is human resources plan was funded by several donors, including the United Kingdom's Department for International Development, and targeted fi nancing for HIV from the Global Fund [17]. Training, incentives and salary top ups were used to recruit and retain health staff [18]. Pre-service training capacity for professional health workers was boosted and medical doctors were imported. Th e role of health surveillance assistants, until then mainly in charge of minimum basic primary healthcare activities, was strengthened. Together with newly created cadres, patient support attendants or expert patients, they now formed a network in the community for treatment adherence support, drug refi ll and defaulter tracing of patients not showing up for appointments. Regular refreshment training and supervision activities were introduced to ensure consistency and quality of care. Additionally, eff orts were made to facilitate access to ART care for health staff [17].
Similarly, in 2003, the Ethiopian Federal Ministry of Health launched a new healthcare plan, the "Accelerated Expansion of Primary Health Care Coverage", to extend quality preventive and selected curative healthcare services to the entire population, with special attention to mothers and children and people living in remote areas. Health extension workers were trained and appointed to each village, and were from now, in charge of 16 packages of basic preventive and selective curative services [19]. As the HIV epidemic worsened, in 2004, the national HIV/ AIDS policy was adapted in line with the existing health extension programme, and health extension workers were appointed for additional HIV-related tasks. By then, the human resources crisis was recognized as a major bottleneck, and a comprehensive strategy for human resources for health was thus developed with the support of the World Bank, the Global Fund and PEPFAR [20][21][22][23]. Th is plan focused on the recruitment and retention of health workers through pre-service training of health staff , introducing incentives packages, shifting tasks to lower-level health cadres, and creating new cadres beyond the health extension workers. Community counsellors, HIV-positive peer educators and homebased care providers were appointed to health services, and were responsible for monitoring and evaluation, ART administration, adherence counselling and defaulter tracing. Th ey link the community interventions to the formal health services. Concurrently, a signifi cant propor tion of the HIV-targeted resources was invested in additional health system strengthening interventions, such as the construction and revamping of health infrastructures and laboratories [20,21].

Methods
Th e pathway displayed in Figure 1 was used to examine the eff ect of ART scale-up interventions on health resources, health service delivery platforms and, subsequently, on general health outcomes and goals. Th is pathway draws upon a health system framework from van Olmen et al ( Figure 2) [9,24], which links the allocation and management of resources (human resources, fi nances, infrastructure and supplies, and monitoring and Th is article focuses mainly on the possible positive spill-over eff ect of ART scale-up interventions on human resources policy changes and the health workforce. Th e principal indicators examined are: health staff availability ( Table 2); the number of newly recruited staff ; the number and type of new lay cadres; and the pre-service training and supervision off ered.
Second, we looked at the impact of health workforce changes on service delivery platforms. We considered both clinical care and community involvement, including ART care delivery. Clinical ART care is defi ned as the delivery of integrated and decentralized ART services at health facility level. Community involvement in ART care relates to tasks performed in or by the community, such as adherence support groups and defaulter tracing. Main indicators are the number of functional health facilities and the number of health facilities available for off ering primary healthcare. Subsequently, we analyzed the eff ects of adapted service delivery platforms on HIV-and non-HIV-related health outcomes and goals.

RESOURCES For support functions
Th e main indicators considered for health outcomes and goals are, respectively: retention in care; the utili zation rate of primary health and maternal and child care services; the maternal, infant and under-fi ve mortality rate; and the TB treatment success rate, as the maternal and child health and tuberculosis services display the most overlap with HIV activities. Th e impact of the other health system resources, i.e., drug supply and monitoring and evaluation, was not analyzed, as both countries implemented parallel systems for antiretroviral procurement and supply and for HIV monitoring and evaluation [21,25].
Key health and health services data on the indicators just described were compiled and compared using several sources: ministry of health documents (national health reports and working plans, and donor documents), evalu ation and activity reports, scientifi c publications on human resources and health systems, and contacts with key informants in the countries. Th ese data were com-

Human resources
Th e change in human resources policies in Malawi and Ethiopia led to an increase in the number of trained health staff at health facility and community level, respectively, by 65% and 302% (Table 2). In Malawi, the reinforcement of pre-service training resulted in a 72% and 12% increase, respectively, in the number of graduated medical doctors and nurses [17]. However, the major increase in the health workforce was noted among community health workers and lay workers. In Malawi, between 2004 and 2009, the number of trained health surveillance assistants increased signifi cantly from 4000   The total number of health workers is composed of the number of medical doctors, nurses, community health workers and other health worker cadres (nurse assistants, clinical offi cers, medical offi cers, lab technicians, etc.), which are not specifi ed in Table 2.
to more than 10,000, and the proportion of health facilities with suffi cient staffi ng available increased from 4% to 13% [17]. In Ethiopia, between 2005 and 2010, almost 34,000 health extension workers and 5500 community HIV lay counsellors and peer educators were trained and employed throughout the country. An additional 3573 health offi cers were trained as mid-level health profes sionals in charge of primary healthcare services, includ ing ART care, and supervision of health extension workers. In a study in Rwanda, this task shifting to lower-level health cadres led to a 76% reduction of doctors' time, freeing up more time for primary healthcare services [26]. Over the same period, the proportion of health workers receiving regular supervision increased by 27% [21].

Service delivery
In both countries, the expansion of the health workforce facilitated the implementation of a public health approach and the progressive decentralization of ART care to the primary healthcare level. Medical and administrative tasks were delegated to lower-level health cadres. Nurses were involved in the initiation of ART and community health workers in adherence support, defaulter tracing, ART distributions and non-HIV-related tasks, such as vaccination, tuberculosis treatment, family planning and antenatal care.
At the same time, this health workforce expansion contributed to an overall increase in functional health facilities. In Malawi, between 2004 and 2010, the proportion of all health facilities providing immunization and family planning services increased steeply, from 9% to 74%. Th e availability of basic emergency obstetric services in health facilities rose from 2% to 65% [25]. In Ethiopia, maternal and child services (antenatal care, obstetrics, postnatal care and expanded programme on immunitization) were progressively decentralized, along with the HIV/AIDS services to the primary healthcare level. Th e number of health facilities providing integrated management of childhood illnesses services increased from 303 in 2007 to 1011 in 2009. Additionally, HIV funds were used to construct and revamp health facilities, increasing their number from 3544 in 2004 to 17,300 in 2010 [20].

Health outcomes
Th e decentralization and integration of ART services in primary healthcare services contributed to a signifi cant scale up of the number of patients on ART in Malawi and Ethiopia. By the end of 2009, respectively, 198,846 and 176,632 HIV-infected patients were on ART [27,28] ( Figure 3).
In Malawi, the proportion of patients initiated on ART in WHO stage 4 decreased from 25% in 2005 to 9% in 2010. Th e 12-month retention rate had increased to 79% by 2009 [27,29] (Table 3).
During the same period, a remarkable improvement in the utilization of key primary healthcare services,

Patients on ART Malawi Patients on ART Ethiopia
Health facilities providing ART care Malawi Health facilities providing ART care Ethiopia including mother and child services, was noted [30] ( Table 4). Th e proportion of children under fi ve years with acute respiratory infections who sought treatment doubled. Th e proportion of assisted deliveries, antenatal care attendance and measles vaccination coverage also improved. Th e increased health workforce and the higher number of available health facilities contributed to a doubling of the primary health capacity in both countries. In addition, the increase in health service utilization rate can be attributed both to the choice to deliver HIV services at primary healthcare level and to the reinforcement of the links between HIV and maternal and child health services. Th e involvement of the community in care delivery encouraged social mobilization and improved access to health education. In Ethiopia, pregnant women in communities with an active health extension programme show up signifi cantly earlier in pregnancy for antenatal consultations compared with those living in villages without health extension workers [20]. A study in a rural district in Malawi showed an eight-fold increase in the annual tuberculosis detection rate after involving com munity workers in the active case fi nding of tuberculosis, and a signifi cant reduction in stigma in the community [31].

The impact on health goals
In Malawi, the early mortality rate after ART initiation dropped from 15% (Quarter 3, 2005) to less than 5% (Quarter 2, 2010) [27]. In Addis Ababa, Ethiopia, a steep decline in population-level AIDS mortality was also reported following the introduction of ART [32].
In both countries, improvements in tuberculosis treatment success rates and declines in maternal, underfi ve and infant mortality rates can be potentially attributed to enhanced accessibility and use of health services (Table 5). A study analyzing the all-cause mortality rate in rural Malawi showed an overall reduction in mortality by 37% over an eight-year period (2000-2008) [33].

Discussion
Th e Malawi and Ethiopia country studies demonstrate how accelerated ART scale-up interventions triggered human resource policy changes, which contributed to increased human resource availability at primary care level, facilitating increased access and quality of primary healthcare and inducing a positive spill-over eff ect on general health outcomes and goals.
In both countries, the shortage of human resources for health was identifi ed as a major bottleneck in ART scale up and was given high priority at all levels, with the adaption and boosting of human resources policies. Government and donors negotiated and agreed to use a substantial proportion of the available increase in HIVearmarked donor funding in this period for general health system strengthening; it was obvious that public health ART care delivery could only be sustained through a reinforced health system. Within health system strengthening, HIV-earmarked funds were mainly used for fi nancing human resources plans, which helped increase the health workforce availability, through better salaries, pre-service training, improved geographical distribution of health staff , and reinforcement of the community-based cadres [34]. In Ethiopia, HIV funds also addressed the lack of peripheral health facilities by constructing additional health facilities [20].
Th e creation and recruitment of new health cadres, in combination with task shifting of medical and administrative responsibilities to lower-level health cadres, catalyzed training and involvement of lower-level cadres in other non-HIV activities. In Malawi, access by health staff to ART signifi cantly reduced absenteeism rates. ART saved at least 25% of health staff lives after 12 months of treatment [35,36]. In both countries, the increased motivation through regular incentives, training and supervision, and improved working conditions led to a better retention of existing health workers.
Collectively, these modifi cations contributed to an increase in the health workforce and the number of functional health facilities, which allowed the rapid scale up of ART services and the better retention of patients in ART care. Yet, this increase also facilitated the accessibility and coverage of non-HIV health services [37]. ART, antiretroviral treatment; GDP, gross domestic product.
One of the major future operational challenges remains developing a model of care able to cope with the evergrowing case load of HIV-positive patients requiring lifelong treatment and retention in care adapted to the resources available. Further, incorporation of this model in the general health system and strengthening of the health system to ensure sustainability and scalability in the future needs to be addressed. Th ere is a potential risk of saturating the existing health facilities, jeopardizing the quality of care. In order to deal with this challenge of achieving eff ective and sustainable ART care at primary healthcare level, community involvement, promotion of awareness and patient self-management are crucial [38,39,40].
Studies in Malawi, Ethiopia and South Africa have documented that the active participation of people living with HIV/AIDS and the community in the planning and provision of care not only facilitated adherence to lifelong ART, but also led to increased patient treatment literacy, empowerment and stronger linkages with the community [41,42,43]. Th is patient empowerment potentially gives more voice to patients to advocate and claim their rights to healthcare. Th is will, in turn, have a positive eff ect on the uptake of health services.Although the response to the HIV epidemics in Malawi and Ethiopia was exceptional and not representative for other countries with high HIV prevalence, some lessons learnt on how scaling up ART off ers an opportunity to strengthen the health system can be useful in other contexts.
Th is analysis points to the possibility of obtaining improved health outcomes beyond HIV through scaledup ART interventions if this process is planned with a comprehensive and long-term vision, involving all actors, especially at community level. Th e country examples show that a human resources analysis with a system-wide view is crucial from the very beginning, when one considers the implementation or adaptation of diseasespecifi c activities. Th e diff erent bottlenecks of the health system and possible spill-over eff ects should ideally already be identifi ed and addressed proactively when planning an ART programme expansion, as was done for the human resources issues in Ethiopia and Malawi [44,45].
In addition, it is important to improve coherence and coordination between the diff erent chronic disease responses, primary healthcare and health system  strengthening eff orts to avoid fragmentation, improve the interface, and adapt policies to maximize synergies. From these case studies, we also learn that to achieve better synergies between HIV disease-specifi c programmes and health systems, several preconditions need to be taken into account, including strong political commitment from governments and decision markers, donor fl exibility and suffi cient fi nancial resources to eff ectively execute planned activities. Th ere are a number of limitations to our analysis. First, the analysis links the positive spill-over eff ect of changes in human resources policy due to funding from global health initiatives to health outcomes. Our results, however, cannot demonstrate a direct link of the human resources policy changes on non-HIV health outcomes due to data limitations and, more specifi cally, the availability of a limited number of indicators measured at two points in time only. To measure and control the impact of HIV funds on non-HIV outcomes, a larger set of health indicators with more measurements over time would be needed.
Second, as health system strengthening is a complex endeavour, with a variety of funding sources, it is diffi cult, indeed, to attribute overall results to one specifi c intervention. Other internal or external contextual factors, such as a better socio-economic situation, a more stable political situation, a higher level of education and an improved nutrition status, might confound these results as they can obviously also boost the general health status of the population. Moreover, eff orts of other diseasespecifi c programmes, free access to healthcare and improvement of primary healthcare services, including maternal and child healthcare, can also lead to better health outcomes. However, our analysis indicates no documented evidence of substantial changes in other factors, programmes or patient user fees for health services in this time period.
Th ird, another limitation in our study is the use of routine data for this analysis, which might be incomplete, including possible recording errors or biases. We did, however, cross check data from diff erent sources and found no major inconsistencies.
More systematic quantitative and qualitative research in several countries is defi nitely required to evaluate the impact of the diff erent ART programme components on health systems and to assess how ART delivery models can contribute to health system strengthening. Th ere is a need to explore the potential use of HIV resources and adapted ART care delivery platforms in the care of non-HIV diseases and the role and the impact of increased user involvement on general health outcomes.
Most developing countries are experiencing an increased burden of non-HIV chronic diseases, such as diabetes and hypertension. Th is burden remains, however, largely neglected in terms of service delivery. Scaling up the ART service model, based on integration in primary healthcare, self-management and community involvement, could have a potential role in the management of other chronic diseases requiring continuity of care [46,47,48].

Conclusions
New additional fi nancing has been invested by global health initiatives for health system strengthening to scale up HIV programmes in sub-Saharan African countries, inspired by a growing awareness that these systems are critical to off er long-term sustainable management of HIV. Our study shows that the scale up of ART services does not adversely aff ect the performance of non-HIV health services. On the contrary, our study shows that the health outcomes of several non-HIV services improved in Malawi and Ethiopia during the same period. Based on both country examples, we conclude that interventions aimed at the expansion of ART services and improvement of long-term retention of patients in ART care can have positive spill-over benefi ts on the general health system.
In both countries, HIV/AIDS-earmarked resources were used to strengthen the health system through tackling the shortage of human resources for health. Increased access to health facilities at community level due to decentralization of services and task shifting to lower-level health cadres and increased involvement of community health workers through community-based networks may have contributed to improved health outcomes beyond HIV.
Th e ART service delivery models can thus potentially serve as an eff ective modus operandi to tackle other common chronic diseases in low-income countries. Th ese models allow countries with relatively weak health systems to cope with the ever increasing number of patients requiring care, to reduce the workload in the health facilities, and to improve retention in care. For this to happen, the diff erent health partners need to acknowledge and capitalize on these interactive eff ects when planning and implementing health activities.
This article has been published as part of Journal of the International AIDS