Social Science & Medicine 129 (2015) 44e50

Contents lists available at ScienceDirect

Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed

A One Health, participatory epidemiology assessment of anthrax (Bacillus anthracis) management in Western Uganda Jeanne L. Coffin a, *, 1, Fred Monje b, Grace Asiimwe-Karimu b, Hellen Janetrix Amuguni a, Terence Odoch b a b

Tufts University, USA Makerere University, Uganda

a r t i c l e i n f o

a b s t r a c t

Article history: Available online 17 July 2014

Sporadic anthrax outbreaks have occurred in and around Uganda's Queen Elizabeth National Park (QENP) for years, affecting wildlife, domestic animals, and humans. Reported outbreaks (2004e2005 and 2010) in QENP collectively killed over 500 wild animals and over 400 domestic animals. A 2011 outbreak in Sheema district temporarily froze local markets while killing two humans and seven bovines. One Health is multidisciplinary at its core, yet studies sometimes focus on the effects of animals on human health to the detriment of investigating the surrounding ecological and cultural contexts. Participatory methods connect problems e such as disease e to their context. A multidisciplinary team used participatory epidemiology and conventional structured questionnaires to investigate the impacts of anthrax on human livelihoods and the related perceptions of conservation, public health, and veterinary health efforts in the QENP area. Proximities to previous anthrax outbreaks and to QENP were treated as risk factors in the collection and evaluation of data. Participants' feedback indicates that anthrax prevalence may be greater than officially reported. Community member perceptions about anthrax and other diseases appear to be more closely related to their proximity to QENP than their proximity to anthrax outbreaks. Neither risk factor had a strong effect on knowledge of disease, nor any effect on behaviors associated with disease response or control. Instead, participants reported that social pressures, the economics of poverty, and the lack of health and veterinary infrastructure highly influenced responses to disease. The complex connections between the social needs and the economic context of these communities seem to be undermining current anthrax control and education measures. This livelihood-based decision-making may be unlikely to respond to educational intervention alone. This study provides a strong base for further research and for improvements in effective disease control. © 2014 Elsevier Ltd. All rights reserved.

Keywords: Uganda Queen Elizabeth National Park Participatory epidemiology One Health Anthrax Conservation medicine Disease management Zoonosis(es)

1. Introduction Anthrax (Bacillus anthracis) is an ancient and virulent zoonotic disease with a poorly understood natural ecology (Schuch and Fischetti, 2009). A vaccine to control anthrax was first developed by Pasteur in 1881 and has been updated (Sterne, 1939). Throughout the early 1900s, vaccination campaigns combined with rigorous movement controls lead to successful widespread disease control (FAO, 2014). As a result, natural anthrax has mostly receded

* Corresponding author. Department of Environmental and Population Health, Conservation Medicine, Tufts Cummings School of Veterinary Medicine, 200 Westboro Road, North Grafton, MA 01536, USA. E-mail address: [email protected] (J.L. Coffin). 1 Present address: 112 Broad Street, Apt. 1, Asheville, NC 28801, USA. http://dx.doi.org/10.1016/j.socscimed.2014.07.037 0277-9536/© 2014 Elsevier Ltd. All rights reserved.

from public attention. However, it is still naturally present worldwide and problematic in many countries and regions (FAO, 2014). Anthrax is caused by B. anthracis, which is a member of the genus Bacillus and part of the Bacillus cereus group. Like many species in the B. cerus group, B. anthracis is a soil bacteria capable of sporulation. This sporulation is an important factor in the complex ecology of B. anthracis (Mock and Fouet, 2011). Its ecology involves many factors including the influence of floodedrought cycles, soil qualities (Ness, 1971; Hampson et al., 2011), other soil flora and fauna (Schuch and Fischetti, 2009; Dey et al., 2012), and transport by insects (Hugh-Jones and Blackburn, 2009). The interaction of these factors leads to long dormant periods between outbreaks. Even in regions of endemicity, several decades may pass between outbreaks (Ness, 1971). Therefore, it is difficult to predict when and where outbreaks will occur. Effective control of anthrax as established by the World Organization for Animal Health (OIE) depends

J.L. Coffin et al. / Social Science & Medicine 129 (2015) 44e50

on cooperation between stakeholders for prophylactic vaccination, vigilance for possible cases, and if necessary, implementation of outbreak control measures (Turnbull, 2002, 2012). Most commonly, herbivorous mammal infection occurs through the ingestion or inhalation of spores contaminating soil or forage. The disease progresses rapidly as spores transform into vegetative cells and multiply, leading to septicemia and death (FAO, 2014). Clinical signs in herbivorous mammals include sudden death, blood oozing from bodily orifices, and occasional lack of rigor mortis (Dragon et al., 1999). B. anthracis sporulation is triggered by contact with oxygen, which is facilitated by decomposition and/or scavenging activity. Unlike the fragile vegetative bacilli, anthrax spores are capable of resisting hostile conditions (Turnbull and Baron, 1996). Recent studies have indicated that B. anthracis, previously thought to remain dormant in soils until entering a mammal, may have other methods of replication. It may replicate as a saprophyte in plants (Saile and Koehler, 2006), symbiotically with soil worms via bacteriophages (Schuch and Fischetti, 2009), or within soil inhabiting amoebas (Dey et al., 2012). These studies help explain how soils remain infective for long periods of time. Human infection occurs through several different routes. Inhalation infection can occur during butchering or curing processes. Infection can also occur through ingestion of contaminated meat. Cutaneous infection can occur during butchering or via insect transport through open wounds (FAO, 2014; USDA-APHIS, 2006). Inhalation and gastrointestinal anthrax are difficult to diagnose early and progress rapidly to septicemia and death. Cutaneous infection can also be fatal, but is more easily diagnosed (Mock and Fouet, 2011). Treatment with antibiotics is effective if diagnosis is timely. The disease itself and the standard control methods have serious economic, biosecurity, and conservation implications. Recently there have been several outbreaks among wildlife in Uganda's Queen Elizabeth National Park (QENP). The 2004e2005 QENP outbreak killed 499 animals: mostly hippopotami, ungulates, and elephants (Wafula et al., 2008). Anthrax was evaluated and designated an “endemic population regulator” for hippopotamus populations in QENP (Wafula et al., 2008). Such a cycle of disease recurrence results in higher infection risk among towns and villages near the park (Hugh-Jones and de Vos, 2002). However, the most recent outbreak in Uganda was in 2011 in Sheema District (more than 50 km from the park) and involved no wild animals. This outbreak temporarily shut down meat, dairy, and livestock markets and claimed the lives of two humans and nine livestock (Promed-mail, 2011; Bagonza et al., 2011). Anthrax is a reportable disease in Uganda. Surveillance is led by the Ministry of Health, the Ministry of Agriculture, and the Uganda Wildlife Authority and involves other stakeholders including nongovernmental organizations, businesspersons, farmers, and politicians. Several disease control methods in accordance with OIE guidelines (OIE, 2012) for anthrax have been reported sporadically to the OIE, including notifiable disease status, general surveillance, movement control within country, targeted wildlife surveillance, border control, and routine vaccination (OIE-WAHID, 2013). However, the regularity of outbreaks within wildlife and concurrent irregularity of outbreak reports in domestic livestock could indicate underreporting of anthrax in Western Uganda. Underreporting could significantly, if unintentionally, impact the control of a real public health threat. One Health (OH) is the concept that animal health, human health, and ecosystem viability are intricately linked. Too often studies using OH focus on the effects of animals on human health without thoroughly investigating the surrounding ecological and cultural contexts. This study used community knowledge to investigate the context of possible underreporting of anthrax

45

outbreaks and factors surrounding anthrax presence. The potential underreporting of anthrax concerns agriculture, trade, and security; this merits investigation. The team assessed the level of anthrax in study areas, drew out the underlying factors of disease occurrence and related behaviors of study participants, and probed participants' understanding and knowledge of anthrax and other zoonotic diseases. 2. Methods 2.1. Study design This study was designed using OH as a framework to leverage interdisciplinary collaboration and knowledge, and selected research methodologies from multiple disciplines. The team included two students from Makerere University's African Field Epidemiology Network (AFENET) fellowship and one student from Tufts Masters in Conservation Medicine program. This team collaborated under supervision of faculty from both universities, as well as staff at the Uganda Wildlife Authority and the International Livestock Research Institute (ILRI). Students and advisors collectively had expertise from veterinary medicine, public health, ecology, wildlife management, and qualitative methodologies. Two main methodologies were used. Firstly, veterinarians trained in epidemiology and public health conducted conventional public health survey methods. Secondly, appropriate participatory methodologies were chosen from social science. The participatory approach recognizes the importance of local knowledge and buy-in to project success and research veracity. It aims to include all stakeholders in the process of problem-solving, development, research, conservation, or disease management (Mariner and Paskin, 2000; Catley, 1999; Chambers, 1994a, b). Participatory epidemiology (PE), in particular, is used in the evaluation and design of disease management (Catley et al., 2012). The multiple levels of stakeholders from health, trade, and conservation anthrax affects imply that engaging affected communities may be particularly important to disease management. The highest level in the typology of participation (Pretty, 1994) shows participants initiating and maintaining project coordination; however, other levels of participation may be also appropriate. This study aimed for a level 4 or 5 out of 7 in the typology. 2.2. Study site selection Kasese and Sheema districts were chosen based on proximity to past outbreaks and to QENP. The location of cases in the 2004e2005 QENP outbreak and the center of the 2011 Sheema outbreak were selected as pertinent “anthrax events.” Anthrax events were defined as an officially reported recent case or outbreak. Kasese is 370 km from Kampala on the western edge of Uganda. Sheema is 300 km from Kampala, and is 100 km south east of Kasese center. Within districts, purposive (risk-based) sampling was used to select subcounties where “near” was 10 km to event or park boundaries. The binary categories of “near” and “far” were chosen based on the assumption that individual livestock, wildlife, and persons mostly stay “close to home” in this region. This is largely true, as livelihoods don't involve nomadism and high population density discourages long distance wildlife movement outside of QENP. Different spatial categories would be necessary in areas where these were untrue. Two villages were chosen purposively from two subcounties (Fig. 1: village locations). Each village contributed one or two focus groups (5e15 individuals), observations from one day, and 8e15 questionnaires to the study. In Sheema district, 29 questionnaires were completed at a farmers' workshop with individuals from

46

J.L. Coffin et al. / Social Science & Medicine 129 (2015) 44e50

Fig. 1. The above maps illustrate the locations of the chosen study villages relative to the defined risks: recent outbreaks in 2004/2005 and 2011 or Queen Elisabeth National Park (QENP). Village names have been given arbitrary letters to help preserve participant anonymity.

many villages. The home villages of this cohort were divided according to the same proximity to risk (“near/far”) for analysis. In total, the team used participatory epidemiology methods with 11 focus groups and 76 structured questionnaires in 7 villages and 1 farmers' workshop. The data were collected over two weeks in July and August 2012.

2.3. Participant criteria To effectively represent local knowledge, individuals were included from multiple social strata, gender, occupation, and adult age groups. Translators were used. Inclusion/exclusion criteria for participants were: (1) Must live in study area more than 50% of the year, (2) must be adults (age 18 and older) of any gender, (3) must consent to participate, (4) must attend the focus group or questionnaire session. Individuals were terminated from the study if they were aggressive and/or threatening, or if they were disruptive to conversation. Participants provided informed consent verbally. The team made participants aware that they could withdraw for any reason, at any time, with no consequence or repercussion. When possible, focus groups were separated by gender. Tufts University Institutional Review Board approval was acquired. Ugandan students followed Makerere University's research ethics procedures.

Ugandan student researchers facilitated the focus groups with the help of translators when necessary. One student always took notes while the other facilitated. First, participants were asked to draw a community map during general discussion of livestock husbandry systems and livestock raising challenges, noting important locations on the maps. If participants mentioned disease as a challenge, diseases were listed and then ranked. Disease importance was defined by each focus group, i.e. prevalence or severity or both, to avoid bias from the facilitators. Proportional piling exercises (Catley and Mariner, 2002) were used (Fig. 2). Impacts of disease on livelihoods were also ranked by importance and discussed. Conversation was also encouraged about zoonoses, anthrax, preferred and appropriate responses to disease in animals, and official partners for disease response. The SSI concluded with feedback about current health systems. Facilitators explained they didn't have the power to make suggestions happen, but that they would report the all findings to local officials. Both Ugandan researchers also facilitated questionnaires with translators as necessary. During focus groups, the note-taking partner also collected observational data. Observations in focus

2.4. Data collection Data collection was based on the principles of participatory epidemiology (PE). Data collection was structured to enable participants to express their perspectives and knowledge with minimal facilitator bias. In discussion, the facilitator aimed to draw out multiple opinions and encourage debate between participants until consensus, or a clear lack of it, was reached. Four key informants e defined as individuals with specific or expert knowledge (Catley and Mariner, 2002) e were informally interviewed during logistical organization to verify the local history of anthrax outbreaks and relevance of study topics. Data collection included three main strategies: (1) Focus groups involving semi-structured interviews (SSIs) and participatory exercises (Appendix 1: SSI checklist); (2) Structured questionnaires; and (3) Observation.

Fig. 2. Proportional piling matrix exercise using beans as counters to indicate ranking and weights of ranks. The above is ranking diseases versus disease impacts.

J.L. Coffin et al. / Social Science & Medicine 129 (2015) 44e50

groups included participant body language and involvement. The team also noted the visual evidence for livelihoods and social structure and comfort of communities during arrival and departure from villages. 2.5. Analysis Analysis began during focus group debriefings, where the note taking and facilitating partners compared their perceptions and notes, cross-checking and clarifying raw data. Post fieldwork data analysis combined three approaches: qualitative, spatial, and statistical. Qualitative analysis involved summarizing main themes and points of each participatory exercise and/or discussion topic. When questions or discussion revealed participant knowledge of disease, specific coding was used to define the quality of knowledge. For example, knowledge of anthrax was judged as good, fair, or poor based on number of correct answers. Poor meant zero to 65% or all correct answers. Similar coding was used for knowledge of zoonoses. Quantitative data from participant rankings in questionnaire responses and discussions were analyzed with basic summary statistics and Chi square or Fisher's exact test. All data from focus groups was triangulated (cross-checked) (Catley and Mariner, 2002) with data from observations, informal interviews with key informants, and/or questionnaires. Spatial analysis investigated differences between focus group responses based on proximity to risk (i.e. anthrax events and QENP). Anthrax events were defined as an officially reported recent case or outbreak. Frequency of responses was compared using the same categories “near” (x
10 km) and “far” (x > 10 km) to risk. Responses analyzed spatially included: 1) livestock rearing challenges, 2) diseases ranked as important, 3) knowledge of zoonoses/ anthrax and 4) accessibility of veterinary care. Spatial analysis was verified using Chi square and Fisher's exact test. 3. Results Focus group participants indicated anthrax occurred more frequently than officially reported in Sheema District, but not in Kasese. Three participant reported anthrax outbreaks in Sheema did not match the official record of anthrax events: one near Kabwohe more than ten years ago in roughly the same location as in 2011; one towards the center of the border of Mbarara and Sheema Districts; and one in southern-central Sheema district. Based on these reports, all study sites in Sheema were within 10 km of a participant report of anthrax, even though study sites I, J, and K were chosen as x > 10 km from an official anthrax event. Table 1 compares the distances of Sheema study sites between participant reported and officially reported anthrax events. None of the participant reports in Sheema corresponded with official anthrax events. However, in Kasese the participant indicated outbreaks did match official anthrax events, and no extra outbreaks were indicated.

Table 1 This table shows the distances between study sites in Sheema District and officially reported outbreaks as compared to distances between study sites and outbreaks discussed during focus groups. Note that distances reported during focus groups are estimates by participants. Study site

Distance to official outbreak

Distance to PE reported outbreaks

F,G,H J,K I

0 km 20e25 km north 20e25 km northwest

~0 km ~5 km west ~4 km east

47

Proximity to anthrax events had little influence on challenges, diseases mentioned, or disease rankings. Table 2 lists the p-values from the chi-squared for independence used to test differences between the groups “near” and “far” from anthrax events. None of these showed a significant difference. The quality of knowledge of anthrax was slightly better “near” anthrax events, but the number of people who had greater knowledge was not significantly different from “far” areas (chi square for independence, p ¼ 0.62). Knowledge of zoonoses was not significantly influenced by proximity to anthrax events either, though it showed a trend towards dependence (p ¼ 0.055). Conversely, proximity to QENP had a strong influence on some categories. Statistical analysis showed that diseases mentioned and ranked were both highly significantly different near the park as compared to far (chi square for independence, p ¼ 0.001). Knowledge of anthrax was significantly better and more common near QENP (chi square for independence p ¼ 0.014). Despite the fact that knowledge of anthrax varied significantly by proximity to QENP, this had no effect on participant response to livestock disease. “Livestock raising challenges” were not significantly different (chi square for independence, p ¼ 0.085) between groups “near” and “far” to QENP. However, there was a qualitative difference: “far” groups all mentioned “disease” and “land shortages.” Among “near” groups, however, a new challenge, “wildlife conflict” was also mentioned. Wildlife conflict is defined as violent interactions between people and wildlife caused by perceived or real incursions by wild animals on crops and/or predation losses from wild carnivores. Participants in Kasese reported crop and predation losses and explained reimbursement programs for such losses are ineffective or non-existent. In proportional piling and ranking exercises, tickborne diseases (e.g. East Coast Fever) were consistently ranked as important, followed by brucellosis. Table 3 visualizes how frequently diseases were given a certain rank in focus groups. Many diseases (including anthrax) had varying rankings from village to village, with no rank clustering. Several diseases had consistently low rankings or were only mentioned by one or two groups. Data from questionnaires showed similar patterns of disease importance ranking. Participants emphasized the importance of common diseases based on regular livestock and revenue losses. Although anthrax is perceived as costly, it is rare, which led participants to rank it as less important in livelihood success. Fig. 3 shows which diseases were ranked of first or second importance. Only three focus groups (two in villages where anthrax events had recently occurred) ranked anthrax first or second in disease importance.

Table 2 Results for chi square test for independence. Many categories showed no difference either near (x
10 km) or far (x > 10 km) from the risk, which was confirmed by the chi square test for independence. Of those that showed a possible trend, only two were significantly dependent, knowledge of anthrax and proximity to QENP, and diseases of concern and proximity to QENP. Knowledge of anthrax and proximity to an anthrax event was nearly significant. Data for proximity QENP and knowledge of zoonoses were insufficient to test. Key:* ¼> suggestive of a trend, ** ¼> significant to p ¼ 0.05, *** ¼> significant to p ¼ 0.001, ^ ¼> no trend/no difference. Chi square test for independence Category tested

Proximity to risk: anthrax event

Livestock livelihood challenges Diseases of concern Knowledge of zoonotic disease Recognize anthrax symptoms Knowledge of anthrax Response to dead/ill livestock

p p p p p p

¼ ¼ ¼ ¼ ¼ ¼

0.077 0.899 0.266^ 0.060 0.055* 0.243^

Proximity to risk: QENP p ¼ 0.085 p ¼ 0.001*** not tested p ¼ 0.062 p ¼ 0.014** p ¼ 0.287^

48

J.L. Coffin et al. / Social Science & Medicine 129 (2015) 44e50

Table 3 Shows how frequently each mentioned disease was ranked by importance. Each “i” represents one instance in which a focus group ranked a disease they had previously mentioned. 11 diseases were mentioned a total of 47 times in 11 focus group sessions. Tickborne diseases were ranked number 1 and Brucellosis was ranked number 2 more frequently any other diseases. Diseases not ranked in Kasese were 3-day sickness, and in Sheema were Trypanosomiasis and worms. Key: ECF ¼ East Coast Fever, FMD ¼ Foot and Mouth Disease, LSD ¼ Lumpy Skin Disease, TB ¼ Tuberculosis. Kasese district

Ranked 1st

Ranked 2nd

Ranked 3rd

4th or lower

Anthrax Brucellosis ECF/tickborne FMD LSD New Castle TB Trypanosomiasis Worms

i i ii i

i ii

i

ii i i i i

i

i

i

ii i

Sheema District 3-day sickness Anthrax Brucellosis ECF/tick FMD LSD Mastitis New Castle

Ranked 1st

Ranked 2nd

Ranked 3rd

4th or Lower

i

i ii

i iiii

ii i i ii i

i iii

i i ii i i

The disease impacts participants mentioned and then ranked as most important were usually economic or social, instead of health related. Participants did mention livestock and human illness and deaths from anthrax, yet they ranked other impacts as more important. Depreciation of the value of the animal or products, cost of treatment, and particularly market closure were all emphasized. The type of social impacts varied by district. Participants from Kasese described increased suspicion from QENP administration of trespassing park boundaries due to disease. Indeed, the observed relationship between QENP and Kasese communities was tense and marked by a lack of trust. Participants from Sheema reported damage to reputation and trade as a disease impact. They felt socially isolated from the other villages and towns during and after anthrax quarantine. People from other communities feared interacting with them or purchasing their animals or products, continuing to avoid the community even after quarantine was lifted. Further, market closures imposed during quarantine damage the livelihoods of communities that experience no infection themselves because they rely on frequent sales of milk, meat, or

Fig. 3. Above are the sums of the times each disease mentioned was ranked of either first or second importance. Anthrax falls between the large group of diseases infrequently deemed important by the eleven focus groups and the two diseases more frequently ranked important, tickborne diseases (especially East Coast Fever) and Brucellosis.

animals for subsistence. As a result, individuals with a suspect case sometimes quickly sell the sick or dead animal at low prices to offset losses and avoid the social and economic costs of quarantine. This reticence to report extended outside of the village. A service delivery stakeholder described intentionally not reporting a case in livestock that tested positive for anthrax. The individual suggested it was best to not report as it would discourage farmers from having animals tested for disease because of subsequent closed markets. Instead, this individual attempted to prevent infection or sporulation by ensuring the suspect carcass was injected with gasoline (petrol) in view of community members and then buried. The stakeholder said the injections made the meat unpalatable and unsalable, removing the incentive to cut losses by selling the carcass at cheap prices. Burying kept carnivores away, mitigating sporulation and ground contamination. Regardless of location, several topics came up. Access to veterinary care and/or medications was difficult, prohibitively expensive, or largely unavailable. The number of veterinarians and extension workers in each district was very low. There were consistent reports of livestock theft and pastureland shortages in both districts. Participants gave many suggestions for solutions. From Sheema: the government should provide annual anthrax vaccination; employ more veterinary officers; provide better education and awareness (“sensitization”) about anthrax and other diseases; and better enforce disease control measures. From Kasese: implement routine vaccination and improve awareness; provide compensation for damages to livestock and crops by wild animals from disease and other causes; increase the number of extension and veterinary officers; construct fencing or other barriers around the park; and increase governmental assistance to improve markets and farming techniques to bolster resilience in times of disease.

4. Discussion Statistical analysis revealed that proximity to QENP had a greater effect on the categories tested than did proximity to anthrax events. In fact, proximity to anthrax made no significant difference in any category, with the partial exception of the slight trend towards dependence for “knowledge of zoonoses.” This may indicate that anthrax events happen too rarely to affect participant perceptions and behaviors. Conversely, diseases mentioned/ranked and “knowledge of anthrax” were found significant (Table 2). This may reflect the effect that park policies, wildlife movement, and tourist presence have on communities. The data indicate anthrax events occur more frequently than officially reported, and reporting is more reliable near to QENP. The difference in reporting between Kasese and Sheema may be due to several factors not studied here: (1) the relative newness of Sheema district may mean reporting structures are still being established; (2) increased governmental presence and activity in QENP; or (3) different economic incentives created by a national park compared to an average community may weaken or reinforce reporting systems. The data from this study suggest the side effects of quarantines and other control measures are major factors in disease reporting inconsistencies, and social and economic ramifications of quarantines are a disincentive to disease reporting. The situation is complicated by inconsistent knowledge about anthrax and appropriate reactions to suspect cases. Continued awareness building (“sensitization”) and education are still needed at the producer and community level to increase consciousness of the appropriate steps to take and how to take them. Since anthrax was ranked as a disease of relatively low importance, using anthrax as a focal point for veterinary or public health education or

J.L. Coffin et al. / Social Science & Medicine 129 (2015) 44e50

interventions may be less effective than using a package of top diseases of concern (i.e., tickborne diseases or brucellosis). However, since respondents of both low and high levels of education indicated they base their decisions on livelihood concerns rather than knowledge of anthrax, further education alone seems unlikely to change behavior. Indeed, despite the fact that knowledge of anthrax varied significantly by proximity to QENP (p ¼ 0.014, Table 2), there was no concurrent effect on participant response to disease. Discussion revealed that participants used alternate strategies. Strategies employed to offset losses and avoid quarantine or infection included quick sales of suspect animals and “over-cooking” suspect meat to “make it safe”. Unfortunately, these strategies do not address infection or ground contamination risks during butchering. Even those with full knowledge of disease risks and the value of reporting preferred to manage risk in their own creative ways e such as burying suspect animals with petrol. In other words, respondents' economic situations and the social effects of quarantine represent real barriers to changing risky behaviors. Participant choices in fact may be appropriate to their situations. They may represent coping mechanisms that effectively balance current economic and health impacts given the options available. However, improving reporting would help Uganda control anthrax more effectively. The finding that the response to anthrax was based on social and economic livelihood concerns rather than ignorance calls into question a typical public health response of education alone. Perhaps providing incentives or financial assistance to communities under quarantine could remove the economic and social pressure from producers and service delivery stakeholders. Or, establishing district-level surveillance for early disease detection and control could redistribute some of the pressures off of village level stakeholders and onto the district. Or, perhaps a strategy that avoids quarantine, as in the “bury with petrol” case, can be developed. Whatever the strategy, the data show there is a need to address the mismatch between accepted public health policies and the livelihood savvy responses of the study communities. This study has several limitations, including the relatively narrow geographic focus, the sample size, and the length of study period. Both the study size and geographic focus limit the generalizability of this study. Two more weeks in the field would have allowed transect walk observations of each village and incorporation of local secondary data. Both data sources would have aided data interpretation, but the described observations and key informant interviews are acceptable substitutes. Ideally, researchers would follow up unofficial reports of anthrax events by going to those locations for further investigation and triangulation, but there was insufficient time. Such follow up could be incorporated in future studies. 5. Conclusions and future directions The strategy of combining One Health (OH) and participatory methods yielded a rich and complex view of a disease and its context. While anthrax negatively impacts the livelihoods and wellbeing of communities in Western Uganda, other factors and diseases also have as strong or stronger effects. When anthrax outbreaks do occur, they impact communities on many levels and lead to economic damage, social distress, and health incidents. There is a lack of education about appropriate behaviors. However, these problems are exacerbated by risky decision-making based on livelihood needs. These findings should be used for further investigation and to inform improvements in disease management. Based on this research, the following strategies might be useful for addressing reporting system weaknesses. (1) Engage

49

communities and other stakeholders when planning and follow up with vaccination programs, education campaigns, and other interventions. (2) Emphasize the OH approach from district to village levels to engage health teams (Kalema-Zikusoka, 2011) in detecting animal disease to increase communication between communities and animal and human health ministries. (3) Investigate areas most at risk for anthrax outbreaks, and establish comprehensive annual vaccination programs there. (4) Pair disease control with poverty management and/or eradication programs supported by the government, national park, or local authorities. (5) Minimize wildlife conflict to build stronger relationships between the QENP communities and local governance. Additional research is desirable to build on these findings and inform anthrax control efforts. For example, assessing the distribution of anthrax in western Uganda and the extent of areas with high risk for anthrax outbreaks would assist targeting interventions, particularly vaccination efforts. Pairing disease distribution data with risk factor distribution mapping could further specify areas most in need. Soil, climate data, and research on real life replication of anthrax during dormant periods could further specify general or possibly annual risk of anthrax infection. All of these study efforts could aid or build on a pilot of effective education, vaccination, and incentive programs to improve future disease control efforts. Vaccination programs are considered an appropriate use of governmental resources when diseases high risk of have the potential for outbreaks (Ramsay et al., 1999; Ndiva Mongoh et al., 2008). Wafula et al. (2008) states that anthrax is a natural population control mechanism for wildlife in QENP. This means that communities near the park will be continuously at risk of anthrax infection. This study confirms that anthrax is affecting Kasese communities. In Sheema district, these data show the disease is poorly documented and that anthrax cases are handled in ways that may contaminate the environment with anthrax spores potentially leading to future outbreaks, as in 2011. Annual vaccination programs are also needed in these areas, but unlike near QENP, more assessment is needed to determine the areas most in need. The full extent of need for annual vaccination in Western Uganda is, at this point, unknown. Part of the principle of participatory methods is working with communities. Communities, especially those near QENP, are experiencing “participant fatigue” from the volume of studies and programs that are active in the area. It is vital to plan in time and resources to ensure that meaningful feedback can be given. This should happen during focus groups and with robust follow up. Ideally, it would bring participants into decision-making for programs, intervention, or even policy. Significant cooperation between organizations, including universities, is vital to follow up success. Further, this could improve data and build relationships into a viable and growing network. Such collaboration would be felt at community level, helping to mitigate participant fatigue while managing diseases like anthrax more effectively. Using an OH framework to modify anthrax control in Western Uganda could improve the efficacy of disease management. In this study, the paradigm of OH and the methods of participatory epidemiology found unexpected potential anthrax reports and shed light on the reasons behind why anthrax may be underreported in Western Uganda. Lack of services, minimal health and veterinary infrastructure, and poverty are all contributing factors that are beyond the reach of typical educational public health interventions. There is a promising opportunity to build on the findings of this study. By utilizing this deepened understanding of the disease context to create more effective, participatory interventions and systems leveraging OH anthrax control could be improved in Western Uganda.

50

J.L. Coffin et al. / Social Science & Medicine 129 (2015) 44e50

Acknowledgments The staff at Makerere University and the Uganda Wildlife Authority provided clearance and guidance. The Tufts Institute of the Environment Fellowship provided most of our funding, and the Tufts Student Chapter of Veterinarians for Global Solutions provided the rest. Other consultants included Dr. Jeffrey Mariner of the International Livestock Research Institute, Nairobi Kenya; and Dr. Gretchen Kauffman and Dr. Elena Naumova of Tufts University. The MCM class of 2012 provided editing and inspiration. Much is owed for the assistance, and much appreciation is expressed. Thank you! Appendix 1 Checklist for semi-structured interviews with focus groups Methods: observation.

community

maps,

semi-structured

interviews,

1. Introduce the appraisal team a. Informed consent b. What are we doing … avoid bias. c. “We welcome questions!” 2. Introduce the key informants/participants 3. Community mapping to capture important features and lay groundwork 4. Major income generating activities a. Livestock species kept b. Husbandry systems 5. Add to mapping exercise a. Grazing locations b. Interaction with wildlife 6. Common challenges a. Identify and describe (at least) three diseases for the area b. Identify main effects of disease on livelihoods 7. Proportional piling exercise on disease impacts a. Ranking b. Matrix e disease impacts vs. diseases 8. Solutions discussion 9. Direct observations (at all times) References Bagonza, A., Busuulwa, M., Akech, S.I., Ngago, B., Nsajju, B., Ohuabunwo, C., Gitta, S.N., Namusisi, O., Wasswa, P., Wamala, J., Ademun, R.O., Mukanga, D., 2011. Multi-species Outbreak of Pulmonary and Cutaneous Anthrax, Sheema District, Uganda [Abstract]. Catley, A., 1999. Methods on the Move. A Review of Veterinary Uses of Participatory Approaches and Methods Focusing on Experiences in Drylands Africa. International Institute for Environment and Development, London, England. Catley, A., Alders, R.G., Wood, J.L.N., 2012. Participatory epidemiology: approaches, methods, experiences. Vet. J. 191 (2), 151e160. Catley, A., Mariner, J.C., 2002. Where There Is No Data: Participatory Approaches to Veterinary Epidemiology in Pastoral Areas of the Horn of Africa (Drylands Issue Paper E110 ed.). International Institute of Environment and Development, London.

Chambers, R., 1994a. The origins and practice of participatory rural appraisal. World Dev. 22 (7), 953e969. http://dx.doi.org/10.1016/0305-750X(94)90141-4. Chambers, R., 1994b. Participatory rural appraisal (PRA): analysis of experience. World Dev. 22 (9), 1253e1268. http://dx.doi.org/10.1016/0305-750X(94) 90003-5. Dey, R., Hoffman, P.S., Glomski, I.J., 2012. Germination and amplification of anthrax spores by soil-dwelling amoebas. Appl. Environ. Microbiol. 78 (22), 8075e8081. http://dx.doi.org/10.1128/AEM.02034-12. Dragon, D., Elkin, B., Nishi, J., Ellsworth, T., 1999. A review of anthrax in Canada and implications for research on the disease in northern Bison. J. Appl. Microbiol. 87 (2), 208e213. http://dx.doi.org/10.1046/j.1365-2672.1999.00872.x. Food and Agriculture Organization (FAO), 2014. Capacity Building for Surveillance and Control of Zoonotic Diseases. Retrieved from: http://www.fao.org/docrep/ 009/a0083e/a0083e00.htm#Contents. Hampson, K., Lembo, T., Bessell, P., Auty, H., Packer, C., Halliday, J., Cleaveland, S., 2011. Predictability of anthrax infection in the Serengeti, Tanzania. J. Appl. Ecol. 48 (6), 1333e1344. Hugh-Jones, M., Blackburn, J., 2009. The ecology of Bacillus anthracis. Mol. Aspects Med. 30 (6), 356e367. http://dx.doi.org/10.1016/j.mam.2009.08.003. Hugh-Jones, M., de Vos, V., 2002. Anthrax and wildlife. Rev. Sci. Et. Tech. De. L Off. Int. Des. Epizoot. 21 (2), 359e383. Kalema-Zikusoka, G., 2011. Integrating wildlife health and community public health to promote conservation and sustainable livelihoods. In: British Veterinary Zoological Society Proceedings. Mariner, J.C., Paskin, R., 2000. Manual on Participatory Epidemiology: Methods for the Collection of Action-oriented Epidemiological Intelligence (FAO Animal Health Manual (FAO), No. 10 ed.). Food and Agriculture Organization, Rome. Mock, M., Fouet, A., 2011. Anthrax. Annu. Rev. Microbiol. 55, 647e671. http:// dx.doi.org/10.1146/annurev.micro.55.1.647. Ndiva Mongoh, M., Hearne, R., Khaitsa, M., 2008. Private and public economic incentives for the control of animal diseases: the case of anthrax in livestock. Transbound. Emerg. Dis. 55 (8), 319e328. Ness, G. B. v, 1971. Ecology of anthrax. Science 172 (3990), 1303e1307. OIE-WAHID, 2013. World Organisation for Animal Health e World Animal Health Information Database. (2014). Disease Control Measures by Country, Uganda [data file]. Retrieved from: http://www.oie.int/wahis_2/public/wahid.php/ Diseasecontrol/measures. OIE B, 2012. Terrestrial Animal Health Code e Anthrax (No. Article 8.1.1). Pretty, J., 1994. Alternative systems of inquiry for a sustainable agriculture. IDS Bull. 25 (2), 37e48. ProMED-Mail, 2011, Nov 8. Subject: “PRO/AH/EDR> Anthrax, Human, Bovine e Uganda (02): (Western)”. Archive number: 20111108.3322. Retrieved from: http://www.promedmail.org/. Ramsay, G.C., Philip, P., Riethmuller, P., 1999. The economic implications of animal diseases and disease control at the national level. Rev. Sci. Et. Tech. Int. Off. Epizoot. 18 (2), 343e356. Saile, E., Koehler, T.M., 2006. Bacillus anthracis multiplication, persistence, and genetic exchange in the rhizosphere of grass plants. Appl. Environ. Microbiol. 72 (5), 3168e3174. http://dx.doi.org/10.1128/AEM.72.5.3168-3174.2006. Schuch, R., Fischetti, V.A., 2009. The secret life of the anthrax agent Bacillus anthracis: bacteriophage-mediated ecological adaptations. PLoS One 4 (8), e6532. http://dx.doi.org/10.1371/journal.pone.0006532. Sterne, M., 1939. The use of anthrax vaccines prepared from avirulent (uncapsulated) variants of Bacillus anthracis. Onderstepoort J. Vet. Sci. Anim. Ind. 13, 307e312. Turnbull, P.C.B., 2012. Guidelines for the Surveillance and Control of Anthrax in Human and Animals (No. WHO/EMC/ZDI/98.6). World Health Organization. Turnbull, P.C.B., Baron, S., 1996. Bacillus, fourth ed.. In: Barron's Medical Microbiology University of Texas Medical Branch, ISBN 978-0-9631172-1-2. Turnbull, P., 2002. Introduction: anthrax history, disease and ecology. Anthrax 271, 1e19. USDA-APHIS, United States Department of Agriculture e Animal and Plant Health Inspection Service, 2006. Epizootiology and Ecology of Anthrax. (accessed on 02.04.12. at) http://www.aphis.usda.gov/animal_health/emergingissues/ downloads/anthrax.pdf. Wafula, M.M., Patrick, A., Charles, T., 2008. Managing the 2004/05 anthrax outbreak in Queen Elizabeth and Lake Mburo National Parks, Uganda. Afr. J. Ecol. 46 (1), 24e31.