Description of the three case studies

The three case studies, on which the framework is based, involved three very different health interventions, with the hope that this variability would provide unique insights into how NTD interventions are negotiated at the local level and the various areas where effectiveness is determined. Following an ethnographic approach, the fieldwork relied on mixed methods, combining quantitative data on coverage, uptake and use of health technologies with substantial and in-depth qualitative research, participant observation, document review and ethnographic notes. For a description of the methods see [21], as well as individual peer-reviewed publications in PLOS NTD [22], Medical Anthropology [23] and Geoforum [24]. For greater detail on the case studies, see these individual publications.

There were five major similarities common to these three interventions that provide for strong comparisons and insights. First, they were all “mass interventions” covering large geographical areas and socio-economic contexts that were financially supported by international donors and planned by technical experts. These projects represented “scaled-up” NTD interventions, aimed at targeting hundreds of thousands of farmers and cattle (Uganda), tens of thousands of dogs (Tanzania) and many hundreds of villages (Zambia). Second, the projects had bold targets that aimed for big impact–as noted in their names and goals, they aimed to “eliminate rabies”, “stamp out sleeping sickness” and achieve “total sanitation.” The assumption was that these interventions would showcase the cost-effectiveness and feasibility of preventing NTDs in rural Africa. Third, they relied on local participation, behavior change and the adoption of prevention practices and technologies, such as latrines (Zambia), veterinary insecticides (Uganda) and dog vaccines (Tanzania). These were, in turn, driven by specific justifications that framed this technology as “appropriate” for rural African contexts: rabies vaccination in Tanzania was free and has very minimal side-effects on dogs; community-led total sanitation (CLTS) is an innovative WASH approach, deemed superior to past subsidy-based sanitation approaches in Zambia because it promised to be “community-led” and reliant on locally appropriate building materials; the Stamp Out Sleeping Sickness (SOS) public-private partnership in Uganda aimed to link agro-veterinary business to sleeping sickness control and livestock improvement by creating new systems of veterinary drug delivery that built on existing practices. These were all “low-cost” and “low-tech” health technologies, underpinning the hope that meeting project targets could be achieved within a short period of time with relatively modest funding. Fourth, the delivery of the interventions were done by district and sub-district actors: government staff, local leaders, extension workers, volunteers and private businessmen. Lastly, specific incentive structures were used to mobilize these people and to motivate them to deliver the intervention and to conduct social mobilization, community engagement and risk communication, with the idea that such approaches would be “locally-led” and “sustainable.”

As with any comparison, there were also important differences. The three interventions involved different types of pathogens–sleeping sickness, rabies and parasitic worms–in three different Eastern African countries with diverse cultures, landscapes, languages, politics and other contextual factors. Similarly, the three interventions all used very different approaches (top-down, participatory and market-driven), institutional arrangements (WHO country office, district teams and a public-private partnership), control technologies (vaccination, social mobilization for pit latrines and veterinary insecticides) as well as delivery networks and local incentive structures to enroll support and participation. These are all summarized in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Important differences between the case studies.

https://doi.org/10.1371/journal.pntd.0006537.t001

Unexpectedly, the individual coverage data found in my large-scale surveys (and validated by my qualitative and observational data) showed that all three of the interventions achieved disappointingly low uptake: 25% vaccination coverage in Tanzania (from a survey of n = 6,157 households), 31% latrine coverage in Zambia (n = 922 households), and an 8.7% market share of the SOS-supported insecticide (known as Vectocid) in Uganda (survey of n = 87 veterinary shops) [22–24]. These projects not only shared these low coverage rates; many of the most salient reasons for why dogs were not vaccinated, latrines were not constructed, and veterinary insecticides were not purchased and used by livestock keepers had many underlining commonalities.

This low coverage was also not inevitable. There were a number of adaptive pathways–as the case studies all make clear below and in the individual case studies [see 22–24]–that could have been used to increase coverage to more acceptable levels both initially and as the interventions progressed over time. Many of these required only modest changes in operational plans, suggesting that the research approach taken here could have substantially improved intervention delivery and population health.

A Socio-anthropological intervention effectiveness framework

In this paper, a framework for socio-anthropological engagement in NTD intervention effectiveness research and program planning is outlined that draws on a synthesis and analysis of these three case studies. This framework seeks to convey the key areas where effectiveness is negotiated as a complex set of interactions between ecosystems, local communities, animals, implementers, health systems and policymakers, within their broader biosocial context. As shown in Fig 1, this has five “effectiveness domains.” Of course, the framework does not intend to be fully comprehensive; rather, it is presented here as a flexible conceptual tool to assist those planning and implementing NTD control to think about these critical issues.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. An anthropological framework for NTD intervention effectiveness.

https://doi.org/10.1371/journal.pntd.0006537.g001

While these three interventions provided the initial analytical lens for the framework, it is worth noting that my thinking has also been informed by subsequent research and control programs, published on mosquito-borne diseases (lymphatic filariasis, malaria and Zika) in Haiti [25], cysticercosis and helminths in Lao PDR [26], rabies in Indonesia [27], cystic echinococcosis in Morocco [28] and other NTDs [29–30]. Overall, this breadth of experience (of different diseases and contexts) has only strengthened my conviction of the utility of such a framework in assisting to inform NTD control interventions.

Domain 1. The terrain of intervention: Seasonality and geographical variation

Global health interventions function over a “local space” where technologies and tools are deployed and issues of coverage and impact are measured and determined. These terrains of intervention are created through strategic decisions based on available resources, political expectations and epidemiological knowledge, and aligned with regional, ecological or district boundaries. There is a specified, and unfortunately often all-too-short, time period where the “field” needs to be understood and transformed [31]. Outside the usual realm of “socio-cultural practices”, diverse social groups, governance histories and micro-ecologies exist within these geographies; local livelihoods interact with land-use patterns, seasonal fluctuations, human movement and migration and new socio-economic pressures, many times outside expectation [32]. My research showed that this variability is important but can easily be subsumed by the challenges of creating technical delivery systems.

First, seasonal and geographical variations in local livelihood systems can be in direct conflict with intervention delivery schedules. Low levels of attendance at CLTS sanitation empowerment meetings in Zambia were due to the program starting implementation when farmers were busy harvesting their crops. Seasonal changes in crop farming and livestock management drove the migration of pastoralists in the Tanzanian rabies case study, the social group with the largest number of dogs. However, the elimination program did not account for this–it was one of the most important reasons for the low vaccination rate. The full-scale (and military/police-led) eviction of thousands of livestock-keepers from the Kilombero Valley after my fieldwork, due to concerns about soil erosion and land carrying capacity, demonstrated just how important population movements can be. Planning effective dog vaccination in subsequent years would require accounting for these communities in the wider WHO elimination area. See [33] for a review of this issue in relation to NTDs more generally.

These dynamics clearly have a major impact on intervention coverage, showing that there is often an optimal “window” for interventions in a given geography, orientated around these seasonal changes and the livelihood patterns of high-risk groups. Geographies need to be problematized as diverse and coupled to livelihood-seasonal change instead of the predominate tendency to conceptualize them as uniform and singular.

Seasonality also influences the purchasing power of households. In the Ugandan sleeping sickness case study, reductions in income during the dry season meant that there was little money available to buy insecticides. This happened to correspond with seasonal tick and tsetse population reductions, mitigating an otherwise dramatic influence of seasonality. But it also meant that the veterinary drug shops established by the SOS program struggled, making it harder for these professionals to make a living. The fragility of local economies and livelihoods in the Ugandan study area were also dramatically affected by a foot-and-mouth disease (FMD) quarantine, an epidemic of cassava mosaic disease, severe flooding followed by drought and a longer-term crisis of land fragmentation, due to high population growth. These vulnerabilities led to a volatile market for the veterinary drug sellers and their community-based animal health workers.

Another, perhaps obvious finding, is that intervention staff have a harder time moving along local terrains during the wet season. The large distances, with poor road conditions, created high translation costs for the SOS veterinarians in Uganda, reducing the time they were willing to educate farmers, at least when direct financial support for community-based education was withdrawn by the financing partners. Large distances also dissuaded the animal health workers from spraying cattle–they did not like “running around chasing people” (as they would say)–one reason, among others, for why they preferred the more profitable work in injectable (treatment) drugs.

Seasonal variation and human movement patterns had an impact on local disease risk perceptions and influenced people’s willingness to adopt health technologies. For example, in Zambia, the fact that people defecated more in their agricultural fields than in their latrines (or around their village) in the rainy season, helped to normalize open defecation and de-motivate households to maintain latrines in the village since construction could only address “half our sanitation problem.” Seasonal weather and changing ecology contributed to many latrines collapsing. Villages that were densely settled, due to historic land-use drivers going back to British colonialism, had created bylaws preventing some households from building latrines due to concerns about contamination and miasmic notions of disease spread.

Generally, the distribution of disease is seldom uniform within a certain intervention area, but often clustered in “hotspots.” Interventions have to make strategic spatial decisions regarding delivery and coverage: the placements of central vaccination point in Tanzania, the location of veterinary drugs shops in Uganda and which villages will be triggered for CLTS in Zambia. In the three case studies, delivery was always conceptualized, by planners, at an abstract district-level without adequately considering the social ecology or epidemiology of the area. In Zambia, certain conditions–many of these tied to ecological characteristics (i.e. access to latrine material) and levels of socio-economic development–offered CLTS the best chance of having an impact on sanitation. In Tanzania, the majority of the dog population was found in very remote pastoralist villages, despite the fact that these were not very well covered by the vaccination program. In Uganda, the SOS supported veterinarians did not focus their efforts on villages with active sleeping sickness patients, despite hospital staff (and hospital records) being available in the local treatment hospital (the focus was on meeting monthly sales targets and no shop was actually located in the most highly-endemic sub-county). Just as epidemiologists now speak about “super-spreaders” [34], intervention planners should operate according to targeted strategies that focus on high-priority areas.

These examples all show the importance of diversifying our idea of the “intervention terrain” as we seek to use knowledge to improve implementation, paying attention to local livelihoods, human movement, seasonal change, geographic variability and epidemiological hotspots.

Domain 2. Social difference and community agency

As with the terrain of intervention, there is also a need to engage with social difference and consider variations in human perspectives, attitudes, logics, social organization, interests and agency. There is a tendency to reify communities as the ultimate target for interventions [35]; but a community is a social construct, a network of relationships and dependencies, and not a geographically bounded unit. Rather, intervention terrains have diverse social groups with differences in wealth, ethnicity, livelihoods, power, knowledge, cultural norms and needs, capacities and constraints. Some people will be more interested than others in the intervention, and more able to adopt technologies and health behaviors. Some will refuse the intervention and may influence others to do the same. Some will recommend changes. This means that planners and implementers need to understand the “public” in public health: they are able to exert influence, make their voices heard and transform plans and policies [36]. But they also have constraints: “throughout the world, those least likely to comply are those least able to comply” [37].

My research found that the intervention adoption process, for insecticides, vaccines and pit latrines, reflected the “uneven playing field” [38] of the rural East African village and deep historical trends of socio-economic and political marginalization. Interventions to alleviate diseases of poverty need to navigate these contexts of destitution, deep inequality and even squalor. The reality is that adoption of health technologies and prevention practices tend to align with higher levels of material wealth and social capital at local level. In Zambia, for example, latrine ownership acted as a symbol of modernity, and it was those households with better education, mobility, access to government subsidies, housing, food security and social networks who build and used them (before and after CLTS). Decades of war, rapid inflation, population growth and other developmental challenges were repeatedly stressed in my Ugandan research to explain why people (sometimes called the “disorganized people”) preferred the cheaper non-tsetse effective insecticides (those only effective on ticks and not the tsetse flies that spread deadly sleeping sickness), despite many knowing that they were inferior products. In Tanzania, canine vaccination was not necessarily influenced by wealth but rather by the motivation for people to keep dogs–dogs that were better cared for and had a defined role in the household were more likely to be vaccinated. Owners were also more able to bring these dogs to the central vaccination point because they were better-behaved dogs. This shows that certain social groups and geographical locations are more likely to be receptive to the intervention.

Motivation to comply with the three interventions, and adopt prevention practices, had a great deal to do with how local people understood the benefits involved. Exposure to “scientific” frames of reference facilitated by veterinary and health extension workers, ideas of social and communal responsibility and a desire to be “modern” were found to play important psychological and cultural roles in all three case studies. Generally, abstract biomedically-defined explanations only went so far in persuading people; rather, what was directly observed, experienced and narrated in locally-understood ways by the community had much more persuasive power. For example in Uganda, many farmers (who were used to spraying insecticides directly onto ticks to kill them) found it hard to understand that cattle sprayed with insecticide could kill a tsetse fly who came to feed on it many hours later. A “hybridism” between experience, local knowledge of disease and biomedical information predominated, something that risk communication and education, which was only done sporadically in these interventions, could have better incorporated.

Local ideas about the value and usefulness of practicing disease prevention strategies were important to people’s motivation, and based on personal experience. In Tanzania, the fear of rabies was frequently mentioned as a motivational force, since people had heard stories (often through their social networks) of previous victims and also feared mass dog culling of non-vaccinated dogs (which local government sometimes implemented on its own accord). The level of input (in time and resources) that people had to invest to rabies control (annual vaccination) was minimal compared to the other case studies, reflecting the lower end of the “participation scale”–as discussed by Rifkin [39]. This certainly helped amplify people’s willingness to comply. In Uganda, experiences of tsetse and ticks and a villages’ location relative to swamps and bushes (breeding areas) were important causes for the level of priority given to sleeping sickness and cattle diseases. Having an animal die from a tick-borne disease was the most often mentioned reason for why farmers began to pay and use insecticides on a regular basis.

De-motivational forces were found to be significant in Zambia and Uganda, where adopting long-term prevention practices were framed as a “gamble” and “risk”, partially because it was never guaranteed to work. This is part of the challenge with adopting health promotion behaviors: they are evaluated based on costs, effort, potential benefits, and in relation to the other multiple vulnerabilities people face. Many simply concluded that the control of neglected diseases was not a priority for them. These were “rare” diseases, and it was better for people to invest their time into something else. This was especially the case when transmission pathways involved more than one route, such as for sanitation-related diseases where one could still get sick despite having a latrine, for example.

An important force behind participation and the promotion of community compliance was the exercise of power, influence and authority at the local level. Punitive efforts included “bylaws” to lock latrine doors and impose “fines” for not having a latrine in Zambia. In Tanzania, this included “village laws” to cull all non-vaccination dogs, dog registration and financial reciprocation laws if a rabid dog attacked someone. Social pressures and simplified local narratives about why people should comply with the interventions were important in normalizing participation and making the case, in locally understandable terms, for compliance and involvement. These included: ticks can kill your cattle; open defecation makes people in the village sick; and not vaccinating your dog can kill your neighbor. These did not necessarily focus on specific diseases, and tended to use stories about local people’s experiences. There was also a different angle to public agency; a household would build a latrine to show their wealth and prestige, to further develop their building techniques and assist in mobilizing (or rebuking) other community members. In this case, those who did not act on these activities were frequently looked down upon, and sometimes mocked and ridiculed as “unsanitary citizens”, even if the ultimate reason was their extreme poverty. Briggs and Mantini-Briggs [40] describe this process in their superb ethnography of the 1990s cholera epidemic among indigenous communities in Venezuela.

Alternatively, all interventions depended on the support and legitimizing labor of local leaders, who were essential in mobilizing community members to attend meetings, spread information and manage aspects of the intervention. In Tanzania, village leadership arranged the location of the central points and disseminated information about the campaign, often door-to-door. In Zambia, poor leadership–where many village leaders themselves, for example, did not have a latrine–contributed to a lack of cohesion and motivation that was only overcome in a few villages with active youth groups, income generation groups and a stronger women leadership culture. A similar case was found in Uganda, where the business aspect of the intervention meant that these leaders only mobilized the community if they were paid a small “motivational” fee.

Domain 3. The strategies and incentives of field staff

Agency, of course, is exerted not only by the recipients of interventions but also by those charged with implementation [25, 31, 38, 40]. The context of the health system, including aspects like human resources, information systems, drug supply chains, basic infrastructure and the culture of management and care, are incredibly important. Interventions require enrolling the support of actors embedded within these systems, and it is often these health staff, community outreach workers and volunteers that are responsible for actual implementation, translating between the different logics and interests of the intervention and the community [29]. In this process, the interests and relationships of these field staff play an overwhelming role in success or failure.

In the case studies, the delivery and planning of interventions was the task of district bureaucrats, extension workers, local leaders, private shops and volunteers. Outside of the training room and models put forth by outside experts, it was these stakeholders (and their working norms, cultural knowledge, incentives and motivations) that shaped the course of events through field-level decisions and social encounters. In Zambia, for example, many of the local champions trained by the CLTS project used the approach in a piecemeal fashion: they did not use the word “shit” (which is used to provocatively illustrate the fecal-oral-disease route) and rarely conducted any follow-up visits with village groups. But they were also clearly not very competent: many were not very well respected by these communities, lacked technical knowledge about latrine construction itself and could not record data properly. In response, people were reluctant to attend community meetings and voiced expectations for hardware subsidies (instead of following the community-driven ideology of CLTS). While local environmental health technicians (EHTs), employed by the Zambian Ministry of Health, were to oversee these volunteers, they were never given any clear expectations, resources or guidance in how to do so. These EHTs themselves operated under severe staff shortages (I often observed clinic cleaners acting as doctors and nurses at these local health centres), corruption, inadequate supplies and low morale.

In Tanzania, veterinary workers were tasked with selecting the central point and mobilizing dog-keepers (in collaboration with village leaders), ensuring adequate supplies of vaccines and administering the vaccines. Although they often enrolled the support of local leaders, I found that many of the vaccine central points were placed in locations close to major roads, far away from the more remote (and difficult to access) areas where, unfortunately, there were more dogs. Information dissemination to these rural locations was limited.

As Pigg [41] has argued in relation to traditional birth attendants in Nepal, the tendency for projects to categorize local actors involved in implementation based on overarching stereotypes can create simplistic assumptions, leading to inappropriate delivery structures. Staff may need to be fired and replaced, making staff quality control and regular team meetings an important component of an effective intervention system.

Many of the volunteers in Zambia had been selected by local political allies with the expectation of gaining monetarily from the program, without doing any serious work. They were provided bicycles and promised money for achieving a set number of Open Defecation Free (ODF) villages but had little oversight, provision of material (pencils, paper and airtime) and clearly defined benefits rewarding hard work. Motivation and support offered to volunteers has also been noted in the ongoing delivery of ivermectin for the control of onchocerciasis in West Africa, where high dropout rates have revolved around a lack of incentives and supervision, long travel distances, other livelihood duties, drug supply problems and working in areas not familiar to the volunteers–volunteers have also been noted to perform better where money has been provided, such as in polio vaccination [42–43].

Field staff have expectations that need to be met to ensure their continued enrolment and performance. For them, global health interventions are a source of work, of salary, prestige and livelihood–as described by Geissler [31] in his ethnographic work in Kenya. The insecticide sprayers supported by the SOS vet shops in Uganda, for example, expected regular workshops, training, subsidies, drugs on credit and various types of free materials such as spray pumps, overalls and gumboots. But these were not provided, at least as often as they assumed would be the case, which reduced their commitment to project goals. Local leaders in Tanzania expected some small financial incentives for mobilizing farmers and livestock keepers in remote areas. In some cases, these were not provided, and coverage in these areas was considered lower than others.

Domain 4. The socio-materiality of technology

When we speak about interventions, we are also talking about technology. The ways that technology characteristics and features embed, and are embedded within, social relationships offers an interesting conceptual approach to explore how global health interventions are delivered. We can call this approach socio-materiality or the “social life” of technology [44–46].

The three case studies showed how the technologies themselves mediated adoption, delivery and use patterns. In Tanzania, rabies vaccines required cold-chain storage that needed to be delivered at specific central points and whose supply depended on international procurement and adequate syringes. Preferences for non-tsetse insecticides in Uganda involved the characteristics of the insecticides and the effect of the drug: the smell, color, packaging, residual period and mode of action. The fact that many farmers only used insecticides to target tick predilection sites was responsible for the continued preference for amitraz products, which are not effective on tsetse flies, only ticks. To use the insecticides, farmers had to buy or borrow spray equipment (which often broke) and needed protective gear. Without them, they used water bottles, which meant that dilution rates and application methods were not ideal.

A different scenario presented itself in Zambia. Weather and insects destroyed poorly constructed latrines while the level of faecal matter in the pit created impressions that they were “unhygienic.” People had to negotiate the landscape as they searched for the more durable logs to construct the latrine base and acquire the necessary material (bricks or bamboo) to make the superstructure; these were increasingly difficult to find in some areas due to land-use pressures and environmental change. Latrine construction was variable, with many different designs. Building was influenced by the technical knowledge of the owner and builder and their relationships. Latrines required maintenance, had different longevities and were spaced away from homes. Access to durable materials and building techniques played a major role in influencing latrine construction, use and maintenance.

Technologies, social forms and ecological characteristics are embedded within a dynamic web of causality. Viewing health technologies as having a socio-material existence promotes understanding these dynamics as an essential step in intervention effectiveness.

Domain 5. The governance of interventions

A true science of global health delivery cannot simply focus on the deployment of predetermined interventions but should also critically evaluate the appropriateness of these interventions, their policy models and wider political economy. Negotiating bureaucratic procedures, knowledge flows, authority structures and power struggles between difference stakeholders in a postcolonial world are inevitable components, with major repercussions for the planning and implementation process. Interventions are also designed through policy narratives that define problems and solutions in specific ways; but these storylines can be overly simplistic to reduce uncertainties, appeal to ideals of feasibility and to enroll support (see the work of Roe [47] for an overview of this position).

For example in Zambia, the larger policy environment of the Millennial Development Goals (MDGs), local government decentralization reforms, the failure of past latrine subsidy approaches and the need to “reinvent” the sanitation sector as well as a global discourse about the appeal of the CLTS technique itself (its simplicity, low cost and impact) motivated the rationale for its implementation. In Uganda, the use of private veterinarians to sustain sleeping sickness parasite reductions, after mass cattle treatments, was generated through an emergency narrative of the eminent merger of the two sleeping sickness forms (Rhodesian and Gambian Human African Trypanosomiasis (HAT), which is unique to Uganda), the synergies between business and public health and the low cost and simplicity of insecticide application. These cogent narratives were framed as locally appropriate but ended up locking themselves into certain delivery pathways involving a specific set of actors, that was difficult to modify–a pattern discussed at length by Leach et al. [48] in relation to ongoing problems with international development planning and implementation more generally. Hence the narrative helped enroll certain actors and perspectives while excluding or marginalizing others.

Policy processes can drive the mobilization of resources and the arranging of intervention strategies in very linear and technocratic ways, where confidence in the intervention’s social engineering (as the political scientist James Scott in his well-known work, Seeing Like a State [49], would say) becomes over-extended. In my case studies, the interventions were funded, managed and driven by different stakeholders–international agencies, academic institutions, the private sector, philanthropic foundations and a variety of government ministries (this institutional ecosystem for NTD control is described in the review paper by [50]). Multiple bottlenecks in the planning, managing and governance of the interventions limited field-orientated pragmatism, most often influenced by various differences between the stakeholders involved and their ability to learn from operational mistakes and maneuver within their spheres of influence.

In Zambia, the focus of UNICEF on strengthening local government decentralization led to CLTS funds and management being channeled through the Ministry of Local Government and Housing (MLGH). However the MLGH rural water and sanitation department had little experience with participatory methods or in rural sanitation. In my study district, the department was staffed by recent graduates from other urban areas of Zambia who had somewhat patronizing views of “dirty villagers”, paid no attention to the past history of sanitation interventions in the district and did not have a strong desire to involve chiefs, EHTs and local volunteers, as emphasized in “true” CLTS field guides. Mismanagement, then, was a key to the low effectiveness of the project, but so were the important institutional histories, norms and conflicts between the different stakeholders. Had the district focal person for CLTS been more committed to success, and been more supported in this regard, outcomes would likely have been very different, as occurred in neighboring districts.

These issues were also encountered in Tanzania and Uganda. The rabies elimination project, centrally organized by the WHO office in Dar es Salaam, distributed equally set budgets to all 28 districts irrespective of geography, infrastructure and dog populations. Kilombero and Ulanga districts, however, were some of the largest districts and a more flexible budget planning approach would have helped address many problems. The fact that the program budget was often sent at unpredictable times of the year, and needed to be used before the end of the fiscal year, obliged the district teams to use the funds when extensive flooding and pastoralist migrations had occurred, as mentioned above. Furthermore, the history of Structural Adjustment Policies (SAPs) (macro-economic reforms instituted by the International Monetary Fund and World Bank in the 1980s as part of a neoliberal agenda) on the veterinary sector in Tanzania meant that staff capacity was sub-optimal, and contributed to negative community perceptions and relationships with local vet extension officers. The top-down methods of planning used by the WHO country office maintained these rigidities.

In Uganda, the interests of the private sector partners meant that the project could not easily adapt to promote a cheaper insecticide. The eventual rolling back of financial support for community education (airtime, money for village leaders, motorcycle repairs and salaries) was driven by the aim of creating self-sustaining businesses. Here was a narrative that sustainable veterinary business could drive sleeping sickness control. But this did not take into account the particularities of disease epidemiology, or the fact that the original spray-team model needed to be adapted. Important changes in socio-demographics and delivery during the course of the SOS interventions were also unaccounted for: a total of 80% of veterinary shops in my four study districts (survey in 2012) had been established since the original business model intervention in 2008. However the changing veterinary drug market, the continued movements of infected cattle into the area and the high amount of non-tsetse insecticides being sold were left unaddressed.

These examples show that policy pathways are hard to change once they are set into motion. A lack of finances, capacity and reflexive management–itself influenced by the conceptual frameworks for action and organizational limitations–all serve to maintain existing courses of action, despite the need for adaptation and fine-tuning on the ground.

There were alternative governance arrangements that could have avoided some of the institutional barriers involved—for example, rabies vaccination managed by NGOs in the Serengeti (Tanzania) were known as more adaptive to local circumstances; the medical sector and International NGOs were both considered more competent to implement CLTS; and not having been tied exclusively to the SOS brand insecticide through private sector partners, or working more closely with the local HAT treatment centers, could have opened up the possibility of promoting cheaper pyrethroid products, and/or tailoring approaches to sleeping sickness endemic villages.

Management inertia was maintained by a lack of accountability, poor monitoring and evaluation systems and an obscuring of the true impact of the program on the ground. This is something that critical medical anthropologists, working in diverse locations around the world, have begun to unpack as they shine their ethnographic gaze on to the bureaucracy and performance of global health itself [51–52]. Intervention ownership was an important factor as well; in all cases, the interventions appeared (and were often spoken about as being) “imposed” by external stakeholders and were time-limited. The short project cycle appears to generate its own psychological burden that negatively influences the drive for success for local managers and field staff: why rock the boat if the project is going to end soon anyways, and it will not necessarily further your career interests? Building up delivery networks and service provision is itself a process that, in many ways, defies the otherwise short-term goals and targets. There are also long-term challenges with capacity, state-citizen relationships, employment and governance that need to also be acknowledged.