Socio-economic impact of Foot-and-Mouth Disease outbreaks and control measures: An analysis of Mongolian outbreaks in 2017

Mongolia is a large landlocked country in Central Asia and has one of the highest per capita livestock ratios in the world. During 2017, reported foot-and-mouth disease (FMD) outbreaks in Mongolia increased considerably, prompting widespread disease control measures. This study estimates the socio-economic impact of FMD and sub sequent control measures on Mongolian herders. The analysis encompassed quan tification of the impact on subsistence farmers’ livelihoods and food security and estimation of the national-level gross losses due to reaction and expenditure during 2017. Data were collected from 112 herders across eight provinces that reported disease. Seventy of these herders had cases of FMD, while 42 did not have FMD in their animals but were within quarantine zones. Overall, 86/112 herders reported not drinking milk for a period of time and 38/112 reduced their meat consumption. Furthermore, 55 herders (49.1%) had to borrow money to buy food, medicines and/ or pay bills or bank loans. Among herders with FMD cases, the median attack rate was 31.7%, 3.8% and 0.59% in cattle, sheep and goats, respectively, with important differences across provinces. Herders with clinical cases before the winter had higher odds of reporting a reduction in their meat consumption. National-level gross losses due to FMD in 2017 were estimated using government data. The estimate of gross economic loss was 18.4 billion Mongolian-tugriks (US$7.35 million) which equates to approximately 0.65% of the Mongolian GDP. The FMD outbreaks combined with current control measures have negatively impacted herders’ livelihoods (including herders with and without cases of FMD) which are likely to reduce stakeholder advo -cacy. Possible strategies that could be employed to ameliorate the negative effects of the current control policy were identified. The findings and approach are relevant to other FMD endemic regions aiming to control the disease.

cacy. Possible strategies that could be employed to ameliorate the negative effects of the current control policy were identified. The findings and approach are relevant to other FMD endemic regions aiming to control the disease.

| INTRODUC TI ON
Foot-and-mouth disease (FMD) is a viral disease that has negative impacts on farmers and stakeholders along the value chain in endemic countries and when introduced into previously free countries. Impacts encompass direct losses that limit livestock production (such as decreased milk production, lower weight gains, decreased fertility and increased mortality mainly in young animals), as well as costs associated with the response to disease or infection (such as treatments, vaccination, movement controls and stamping out) (Knight-Jones & Rushton, 2013). Countries with endemic FMD are denied access to some potentially lucrative export markets for livestock and animal products, giving governments a clear incentive to chain resources to control the disease. It is often assumed that by controlling the disease and acquiring the World Organization for Animal Health (OIE) 'free without vaccination' status, all animal holders (regardless of the production system, size, and access to markets) would benefit, either by increasing their income or increasing availability of animal-source food (ASF), such as milk and meat, in the household (FAO, 2011;FAO & OIE, 2012). However, the benefits of controlling the disease in low-and middle-income countries (LMIC) are complex and not well quantified (Knight-Jones, McLaws, & Rushton, 2016;Knight-Jones, Robinson, et al., 2016;Limon et al., 2017). Furthermore, the impact of FMD and consequences of the control programmes on animal holders' livelihoods and food security is rarely explored. Studies have focused on quantifying the impact of the disease in mixed crop-livestock systems in Africa and Asia, and large-scale commercial or pastoral systems in Africa (Jemberu, Mounts, Woldehanna, & Hogeveen, 2014;Lyons, Alexander, et al., 2015;Nampanya et al., 2015;Perry, Gleeson, Khounsey, Bounma, & Blacksell, 2002;Young, Suon, Andrews, Henry, & Windsor, 2013).
However, the indirect impact of FMD control measures in settings where animal holders' diet is based on ASF has not been assessed.
Mongolia is a large landlocked country in Central Asia, bordered by Russia to the North and China to the South, East and West households owning livestock, and a quarter of people employed in the agricultural sector which is dominated by livestock production (Erdenesan, 2016). Livestock are raised predominantly by nomadic herders with production based on traditional herding practices.
Herders move location each season and are typically placed kilometres apart from one another leaving them very geographically isolated.
The rural Mongolian diet is influenced by the extreme continental climate, isolation and nomadic lifestyle, and it is heavily reliant on animal protein and fat (dried meat mostly in the winter and meat and dairy products during the summer) (Jamiyan, 2017b). Although alternative staples and vegetables have been introduced over the years, these are mainly consumed in urban areas. Food security is a key component of the aim to achieve 'zero hunger' (one of the United Nation's sustainable development goals) and has four core dimensions: availability, access, stability and utilization (UNDP, 2015;World food summit, 1996).
In rural Mongolia as in other LMIC, food availability, access and stability depend to a great extent on household-level production, which can be compromised by suboptimal animal health or abrupt changes in herd structure (Rushton, Thornton, & Otte, 1999). In recent years, Mongolia has implemented a series of programmes to reduce food insecurity and improve the nutritional status of the Mongolian population, with targeted social programs to reduce vulnerability to seasonal food shortages (Jamiyan, 2017a).
Mongolia has a very long and sparsely populated land border which makes it vulnerable to transboundary animal diseases. In the past 5 years, Mongolia has been affected by peste des petits rumi- Reports of FMD in Mongolia increased in January 2017 compared with previous 12 years ( Figure S1); with outbreaks of serotype O affecting 810 herders between January and December 2017 in 8 provinces in the Eastern part of the country. Species affected included cattle, sheep, goats and camels. Up to 9 outbreaks (defined as all herders affected during the same quarantine period following outbreak investigation) were reported each month, with more outbreaks reported in the summer and winter periods compared to spring and autumn periods ( Figure S2). The current national FMD control strategy in Mongolia consists of vaccination twice a year in high-risk areas, modified stamping out (i.e. only destroying animals with clinical signs) and movement controls.
Following a report of an animal with clinical signs suspected as FMD, a 10 km quarantine zone is put in place and an outbreak investigation begins. The size of the quarantine zone might vary depending on the location and natural barriers. Animals with clinical signs are destroyed and farmers receive compensation worth 90% of the commercial value of the animals culled. Once the last animal is destroyed, a 28-day quarantine period begins. During the outbreak investigation and quarantine period, no animals or people can move in or out of the quarantine zone. Once the quarantine period has concluded, the subsequent appearance of clinical signs of FMD in a herd is considered a new outbreak and a new investigation and quarantine period initiated.
Estimating the impact of animal disease and resultant control measures at both the national and herder levels provides information that can be used to guide future control policy and resource allocation for animal diseases. This study addresses the impact of FMD in Mongolia in 2017 by (a) assessing the socio-economic impact of FMD and the control measures on herders; and (b) estimating the national gross economic losses during 2017.

| Study settings
This study was conducted in eight Eastern provinces in Mongolia ( Figure 1c). Mongolia is divided into 22 provinces (commonly known as aimags), 335 districts (also known as soums) and 1,800 sub-districts (also known as bags) which are the smallest administrative unit.

| Study design
Using official outbreak reports from the State Central Veterinary Laboratory (SCVL), ten herders affected with FMD were randomly selected in each affected province from outbreaks starting between January 2017 and December 2017 ( Figure 1c). Five additional herders that were not affected but within these quarantine areas during the same period were also selected. Herders affected with FMD in January 2018, but during an outbreak commencing in 2017 were included in the sampling frame. For simplicity, selected herders affected by FMD will be referred to as 'affected herders' and herders not affected by FMD but within the quarantine zone will be referred to as 'quarantined herders'. For each selected herder, the aim of the study was explained and verbal consent to participate was obtained.
If the number of affected herders in a given province was less than ten, all affected herders during the study period were surveyed.
Similarly, if there were fewer than five quarantined herders, all available herders were visited.

| Data collection and questionnaire design
A standardized questionnaire was designed using an exploratory sequential approach (Creswell & Plano Clark, 2011). An initial questionnaire using a combination of closed and open-ended questions was designed based on authors' (GL, NAL) experiences and discussion with members of the SCVL. This questionnaire was piloted in four affected herders from one province (Tuv). Answers given by these four herders were discussed among the team members and used to modify the initial questions according to the local context TA B L E 1 Data collected as part of the herder survey and used for analysis

Herds included in the study and epidemiological characteristics
Descriptive statistics were generated stratified by province, FMD status (affected herder versus quarantined herder) and species.

Socio-economic impact on herders
The following parameters were estimated in order to assess the

Impact on livestock assets
The  Table S2). When assessing the impact due to culling, we considered two scenarios: before herders received compensation and after herders received compensation, which is 90% of the market value according to the regulations at the time of the survey. For those herders that had bought animals to replace those lost during the outbreak, we estimated the difference between prices paid and market price given by MSIS. Finally, for those that had not been able to buy animals the reasons for not being able to replace them are described.

Impacts on herders' income by foregone sales
The proportion of herders that were planning to sell animals or animal products but could not because of the outbreak and the control measures in place were estimates stratified by status (affected versus quarantined). For those that were not able to sell,

| National gross economic losses
National-level gross losses due to reaction and expenditure in 2017 were estimated based on governmental data in a deterministic model. Reaction expenditure: • Vaccines and vaccination included costs of the vaccines and delivery in order to vaccinate all susceptible animals (cattle, sheep, goats, camels and pigs) in the affected provinces.
• Diagnostics included diagnostic costs and laboratory consumables for testing samples from animals with clinical signs.
• Outbreak investigation and surveillance included the costs of staff attending outbreaks, the use of personal protective equipment and collection of samples.
• Compensation for culled animals (90% market value) based on the market price for live animals in each province (Table S2) • Quarantine implementation included cost petrol, per diems for field vets and disinfectant per province.
Production losses.
• Mortality (due to deaths) based on the market price for live remaining 10% was incorporated into the national and provincial cost estimates of 'production losses' which also included the costs of animals that died for which farmers did not receive compensation. These costs were stratified by species (cattle, sheep, goats and camels) and age (0-1 years, 1-2 years and >2 years). No other production losses were considered because affected animals were either culled or died.
The estimate attempted to capture the gross losses during the outbreak and subsequent control measures. Losses over time as a consequence of the outbreak were not considered. All the other costs (vaccines and vaccination, diagnostics, outbreak investigation and surveillance, compensation for culled animals (90% market value) and quarantine implementation) were combined into 'reaction and expenditure' at national and provincial levels. with other species). The most common reason given for keeping animals was as a source of meat for home consumption (99.1%), followed by selling live animals to generate income according to need (82.1%), producing milk (cattle, sheep and goats) to be consumed at home (78.6%) and slaughtering them at home and selling the meat (67.9%).

| Characteristics of herders included in the study
Less common reasons for keeping animals were to produce milk and sell it (29.5%) and to sell animals on a regular basis (29.5%).  Figure S3).
Forty-five herders (40.2%; 26 affected herders and 19 quarantine herders) mentioned they did not know or understand the reasons for the implemented control measures of FMD, and most of those that understood the reasons had a veterinarian as a family member.

| Epidemiological characteristics of FMD outbreaks
Out of the 70 affected herds, the most common month for herd-  Figure S2). The median herd-case duration was 6 days, ranging from 1 to 48 days with no significant difference between provinces (p = .15). Median attack rate was 31.7% in cattle, 3.8% in sheep and 0.59% in goats. The attack rate was higher in bovine calves than adults with the opposite trend reported in sheep and goats ( herds. There was no mortality reported in affected sheep and goats.

Impact on livestock assets
As part of the official control measures, animals with clinical signs of FMD were destroyed. In the majority of herds, the number of animals TA B L E 2 Herd and flock sizes on the day of the survey stratified by province and FMD status. .71 238 (170-425) .36 273 (174-363) .22
At the time of the survey, most herders (92.9%) reported that they had not been able to replace all animals that were culled as part of the control measures, while the rest (7.1%) had only been able to replace some of the animals lost. The prices of these replacement animals ranged from ₮200,000 to ₮1,500,000 (median ₮325,000; US$132.01) in cattle, between ₮40,000 and ₮70,000 (median

Impact on herders' income due to forgone sales
Reduction in the sale of animals was reported by some herders. Half of the affected herders and a third of the quarantined herders were not able to sell animals or animal products that had planned to sell and would have been sold if the outbreak had not occurred (Table 4).
For those herders that were planning to sell live animals, the median income foregone from sales of live animals (during the outbreak and control measures) was ₮8,406,500 (US$3,414) for affected herds and ₮5,003,500 (US$2,032) for quarantined herds, while for those planning to sell milk the median income forgone was 1,200,000 (US$487) for affected herds and ₮3,000,000 (US$1,218) for quarantined herds (Table 5).

Impact on herders' expenditure
Among both affected and quarantined herders, almost half of herders interviewed (n = 48; 42.8%) did not apply any measures to protect non-affected animals beyond the government enforced quarantine.

TA B L E 4 Impact of FMD control measures on herders' livelihoods and food security in Mongolia
Note: Data were collected between May and June 2018 (n = 112). Difference between affected and quarantine herds were assessed using Fisher's exact or chi-squared tests for categorical variables and Kruskal-Wallis tests for continuous variables. a From those that had planned to sell animals or animal products.
using an alternative approach such as fumigating with some plants, washing with disinfectant or moving to another location. No significant differences were found between affected and quarantined herds on the application of control measures to animals that were not affected. Median treatment cost was higher for quarantined herds ₮82,000 (US$33.) than for affected herders ₮18,000 (US$7.3) (p = .04). For herds that treated with antibiotics and reported the length of treatment and cost (n = 4; 3 affected herds and 1 quarantined herd), the treatment lasted between 1 and 7 days (average 3.3 days) and the overall cost ranged between ₮13,000 and ₮20,000 (US$5.28-US$8.12) per herd.
Seven (10%) affected herders and three (7%) quarantined herders purchased milk. Seven (out of ten) provided prices of milk purchased, the median cost of milk reported was ₮40,000 (US$16.2) per litre (min ₮24,000 max ₮225,000), with quarantined herders spending more than affected herders (

Impact on herders' food availability and access
In terms of food security, household (physical) food access and availability were reduced in both affected and quarantined herders and their families. More than two thirds of herders (and herders' families) did not drink milk for a period of time (median 45 days for both groups, range 7 to 120 days in affected herders and 21 to 90 days in quarantined herders; p = .77) or had to reduce milk consumption (median 30 days for affected herders-min 14, max 60 days, and 21 days for quarantined herders-min 14, max 90 days; p = .48).
Milk and meat consumption before or after the outbreak were not recorded, and therefore, differences in consumption at differences points in time were not estimated. Empirical observations suggest that reduced consumption among quarantined herders is due to the belief that consumption of animal products might be unsafe while in the quarantine period regardless if their herd was affected or not.
Notably, the median periods without drinking milk and reduced milk consumption were longer than the mean herd-case duration. In both groups, a sixth of herders and their families did not eat meat for a period of time and a third reduced their meat consumption for at least 20 days (Table 5). Informal observations suggest that food substitution would be low given the nomadic lifestyle of herders and limited availability of other sources of food such as crops although food substitution was not systematically recorded.
Examining milk and meat consumption more closely, herders that reported having stored dried meat (i.e. had some meat savings to cushion them) were less likely to stop (p = .04) or reduce (p < .001) meat consumption for a period of time. Province was not included in the multivariable analysis as the model failed to converge. For affected herders, herders were significantly more likely to have their meat consumption reduced when the herd-case started in Note: Data were collected between May and June 2018 (n = 112). a Only herders that were planning to sell animals or milk were considered.
b 10 reported to buy milk but only 7 gave prices. c Only animals that spent in treatment were considered.

TA B L E 5 Monetary impact of FMD control measures on herders in Mongolia
September or October (i.e. before the winter) compared to those affected in January, even after adjustment for storage of dried meat in the household (Table 6). No statistically significant patterns were found between milk consumption and time of the year when the herd was affected. (Tables S3 and S10). Time of the year when the quarantine was put in place was not recorded, and therefore, the effect between months when quarantine started could not be assessed on quarantined herders.

Other impacts on herders' livelihoods
Borrowing money was a common coping strategy, representing a household income shift from livestock to use of credit. More than half (57.1%) of affected herders and more than a third (35.7%) of quarantined herders had to borrow money for reasons including buying food, buying medicines for family members and paying bills and bank loans. Affected herders were significantly more likely than quarantined herders to borrow money for buying food (p = .008) and/or buying animals (p = .024) ( Table 4). The places that herders borrowed money from was not systematically recorded.
Other negative impacts on herders as a consequence of the control measures included not being able to receive visitors, afford school fees, school closures and household members not being able to return home (Table 4).

| National level
Provinces reporting clinical cases are illustrated in Figure 1c. The number of animals culled, attack rates and mortality rates stratified by province and species are presented in Table 7. The majority TA B L E 6 Final multivariable logistic regression model for identification of factors associated with reduction in meat consumption during the FMD outbreak and control measures in place Note: Univariate models are presented in the supplementary material. The majority of the cost (₮17.7 billion, 96.4%) was due to reaction and expenditure (Figure 2), from which ₮10.4 billion (59%) was due to vaccination: ₮4.4 billion (25%) due to compensation, ₮2.8 billion (16%) due to quarantine cost and the rest (0.86%) due to diagnostic and surveillance. The costs were highest in Sükhbaatar province which made up 30% of the total national cost.

| D ISCUSS I ON
The study provides the first assessment of the impact of FMD con- The median attack rate in cattle in this study was lower than in other endemic settings (Jemberu et al., 2014;Lyons, Alexander, et al., 2015;Vu et al., 2017), which could be attributed to the production system, virus strain, differences in level of immunity and the (modified) stamping out policy in place.
Detailed assessment of the attack rates revealed a high variation in attack rates between provinces although there was no strong statistical evidence to support this observation. Some of this vari- Our study showed that the current FMD control measures have important negative consequences for Mongolian herders. Some herders and their families went without drinking milk for longer than a month, with affected herders up to four months and quarantine herders up to three months. This suggested that the negative effects can last longer F I G U R E 2 Provincial level costs of FMD outbreaks in 2017 in Mongolia represented as production losses and due to reaction and expenditure than the herder-case duration and the official quarantine period. A similar effect was observed with meat consumption with some herders having a reduced meat consumption for up to a month in the case of affected herders and up to two months in the case of quarantined herders. Crucially, the livelihood and food security of all herders within a quarantine zone, including those without clinical disease in their herds and flocks, was seriously impacted. Although the negative impact on herders with animals showing clinical disease can be expected, the collateral damage to farmers that fall in the quarantine zone is usually ignored and has not been previously quantified. In interpreting our results, it is important to consider that these parameters are herders' estimates and recall or reported bias cannot be excluded. The study may have been considered an opportunity to obtain support, and as a result, the negative effect on meat and milk consumption might have been overstated. Longitudinal studies to capture differences on consumption and management practices (such as selling as slaughter rates) during the year and over time should be conducted in the future to better understand patterns without the disease or control measures.
Nonetheless, the information recorded and reported here is valuable baseline information that illustrates the negative impact and can be used in further studies. Some quarantined herders reported avoiding the consumption of animal products from their farms as they perceived them as unsafe to consume, highlighting some of the misunderstanding that might arise during disease control programmes and the need for effective communication among stakeholders at different levels.
An important seasonality effect was observed, with herders' meat consumption more likely to be affected if the outbreak happened before the winter. This outbreak timing is likely to have a greater negative impact on the ability to store dried meat which is commonly done in Mongolia in preparation for hostile winter conditions. However, month when animals are affected (i.e. month when herd-case starts) and having meat stored only explained a third of the variance and so other factors that were not recorded in this study might also play a role in reducing meat consumption. Furthermore, the wide confidence intervals observed in some of the months reflect the variation on the data and the relatively small sample size. Outbreak timing (or another shock situation) has been identified elsewhere as an important factor linked to food stability (Limon et al., 2017)-a food security component that highly depends on the resilience of the household to cope with adverse situations. In contrast to subsistence farmers in other parts of the world, where production diversification is part of farmers' strategy to deal with variability in production (Ellis, 2000;Randolph et al., 2007), herders in Mongolia rely almost entirely on livestock for food and financial security with very limited options for food substitution (Jamiyan, 2017a), reducing their resilience in the face of high impact disease outbreaks.
As a result, available coping strategies to deal with food scarcity are limited, resulting in the majority of the herders incurring extra expenses or accruing debt. For children, not only was their food security compromised but there were further negative effects through temporary lack of schooling and absence of family members. Formally quantifying food substitution and available options for borrowing money should be explored in more detail in the future to better understand the impact and safety nets that herders might have available.
The unintended negative effects of the current FMD control policy in Mongolia identified in this study is at odds with the sustainable development agenda (United Nations, 2015) and the Mongolian national policy to tackle food insecurity and malnutrition (Jamiyan, 2017a, b). However, the research has also identified ways of ameliorating the negative effects. For example, enhanced food support for affected and quarantined herders in parallel with the current FMD control strategy, especially if the outbreak happens in the months before winter, to safeguard nutritional needs of herders and their families. A limitation of this study is that food substitution and food utilization, specifically repartition of food available within the household, was not assessed. In addition, data to capture the perception by gender and differences on how men and women were affected were not collected. It is therefore possible that the extent to which food security is compromised among different household members might differ. Further studies should investigate this and the nutritional implications among different household members, especially pregnant women and children. Macro-level evaluation of food security (market stability) was beyond the scope of this study. Similarly, effect on market prices due to changes in supply and demand of live animals was not estimated. Herders planning to sell livestock during the outbreak or quarantine period may have delayed the sale rather than foregone revenue completely. Therefore, the estimated loss is the gross loss and could be an overestimate. The effects of the current policy over time should be evaluated and quantified in the near future to get a better understanding of the overall collateral effects and benefits.
Most of the herders (94.3%) interviewed had not received compensation at the time of the study, resulting in direct negative impact on herders livelihoods by decreasing herd size and production and increase herders' debt and expenditure. Furthermore, half of the herders did not know or understand the reasons for the control measures and there was some misunderstanding regarding the safety of milk. Similar situations in other countries have led to lack of trust in the veterinary services and non-compliance (Elbers, Gorgievski, Zarafshani, & Koch, 2010;Limon et al., 2014;Smith, Bennett, Grubman, & Bundy, 2014). Concerns about herders reporting new cases and fully implementing animal movement restrictions were highlighted in a previous risk assessment identifying strength and weakness of the Mongolian FMD control system (Wieland, Batsukh, Enktuvshin, Odontsetseg, & Schuppers, 2015). Hence, providing compensation in a timely manner, putting in place procedures to reduce the negative impact of the control measures to herders and improving risk communication should reduce the negative impact on herders' livelihoods and maintain institution credibility.
The estimate of gross economic loss at national level was US$7.35 million, which equates to approximately 0.65% of the Mongolian GDP. Although the current control policy had reportedly controlled sporadic outbreaks in previous years, the outbreaks in 2017 were much more widespread with greater numbers of herders and animals affected ( Figure S2). It is important for any country implementing disease control measures to evaluate the policy against defined objectives to ensure optimal use of resources. Figure 2 shows the expenditure on reaction far outweighs the production losses which is expected with the current control policy. Similar results have been reported in other FMD outbreaks (Thompson et al., 2002). Only the gross costs of the current control policy (within one year) were estimated, and the benefits of indirect costs on reducing the overall impact over time were not assessed. Moreover, a net estimate that incorporates the cost of keeping animals (e.g. feed) was not included. Simulation models to evaluate different control scenarios in order to determine the most likely cost-effective policy should be done in the near future. The model developed in this study could be extended to a full cost-benefit analysis which can be used to inform policy.

| CON CLUS ION
This study described and quantified the unintended consequences of FMD control measures on herders' income, extra expenditure and debt, and food availability and access; and estimated the nationallevel gross losses attributable to the outbreak in 2017. Thorough analysis of our findings has revealed possible strategies that could be employed to ameliorate the negative effects of the current control policy.

ACK N OWLED G M ENTS
The study was funded by the Veterinary and Breeding Agency in Mongolia, Pirbright Impact Acceleration Account projects

CO N FLI C T O F I NTE R E S T
The authors declare no competing interest. Co-authors working for the State Central Veterinary Laboratory or the Veterinary and Breeding Agency in Mongolia helped to provide and coordinate data collection but did not have a role in the analysis and interpretation of the data.

E TH I C A L A PPROVA L
Ethics approval was obtained from the State Central Veterinary

Laboratory in Mongolia and the Social Science Research Ethical
Review Board at the Royal Veterinary College (URN SR2019-0145).
All activities were conducted in accordance with protocols approved by this board.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.