Session I: Measuring animal movements and drivers for FAST disease risk mapping

Recorded Session

Aim: to scrutinize how animal mobility drivers and network analysis can contribute to risk mapping for FAST diseases and how animal movement changes and related risks can be monitored and forecasted.

Main issues:

AGENDA

8 DEC

SESSION I
  •  
Chair
Prof. James Wood
 
Round table moderators
Prof. James Wood
Dr. Bouda Ahmadi
Dr. Corissa Miller
Dr. Paolo Motta

R.R. Kao

G. Chaters, W. de Glanville, L. Matthews, J. Nyarobi, E. Swai, S. Cleaveland, & P. Johnson.

The buying and selling of livestock represents a substantial contribution to the income and livelihood of the people who husband them, however these movements also represent a risk of infectious disease transmission. While the health and commercial costs of disease means that there is incentive to control them, those affected by these costs are often not those most responsible for causing them. Thus market forces cannot be relied upon to restrict disease transmission, and policy-based interventions to prevent pathogen spread are necessary. Modern data-driven approaches to livestock management mean that in many higher income countries, data on livestock movements are collected routinely and stored digitally, and so they can readily be used to inform disease control efforts. However in Tanzania, as with many other LMICs, data on movement patterns must be generated. Here, I describe the collection and analysis of a substantial dataset generated from paper records, which is used to generate a ‘synthetic’ dataset that replicates key features of recorded patterns of movements to represent livestock movement patterns across northern Tanzania. A combination of network analysis and simulation is then used to determine key locations that can be targeted for control of many diseases including the important zoonosis, Rift Valley Fever (RVF). We show that targeting control efforts at locations that are critical because of either their role in the network of cattle movements or the inherent local risk of transmission can provide similar gains in control efficiency. While further investigations are necessary to corroborate these results, they suggest that multiple levers may be available to manage this and other livestock diseases. Efficient control of livestock diseases in areas with limited data and resources available is essential and here, by demonstrating the potential utility of livestock movement data in targeting interventions, this provides an argument for routine collection of such data. This will be especially important for ongoing assessment of risks as we face a future of increased uncertainty in living with environmental and land use change.

Summary presentation
Soon available
Share on facebook
Share on twitter
Share on linkedin

 

 

Andrea Apolloni

Alexis Delabouglise
as representatives of the AQCR
team CIRAD

CIRAD-UMR Astre, France

We are going to present some of the results from our recent works on animal mobility in 2 contrasted Regions: West Africa and South-East Asia. In both cases we tried to understand what the driving factors are behind observed mobility patterns. Different methodologies where used in the 2 contexts.

 

 Live animals trade is in most of the West and North African countries one of the main economics activities. In general, the consumption and production areas are several hundred km apart. Due to the absence of infrastructures, animals are sold alive at local markets to traders, and then moved to capital or coastal cities where they are slaughtered and butchered. Because of this, livestock mobility is an intrinsic component of the farming systems in the region, aiming at optimizing the availability of natural resources (grasslands, surface water) which shows a highly seasonal pattern for a given grazing area. Livestock mobility in the region is a complex phenomenon involving several temporal (from days to months) and spatial scales (from a few km to reach local markets, to international transhumance or trade movements). 

The possibility of providing a reliable picture of livestock mobility in the area is hindered by the fact that few quantitative data are collected.

We are presenting the results of the analysis of data provided by Veterinarian services in West and North Africa countries on ruminants’ mobility. We used complex network approach to describe mobility patterns and a suite of models (gravity, ERGM) to understand its driving factors of animal mobility. The analysis shows the existence of largest transboundary communities of movement, that could facilitate the spread of the disease in Region. The analysis has shown the existence of 2 important periods for mobility: a routine one, and the period around the religious festivity of Tabaski (during which a young male sheep is slaughtered in each family).  The mobility drivers include environmental factors (conditioning the availability of natural resources), commercial reasons (demand and market price), economical (gdp difference between producer and consumer areas) and social factors like the Tabaski celebration.

 

In several countries of Southeast Asia, the control of major infectious diseases of poultry is hampered by the complexity of the poultry trading networks connecting millions of small-scale producers to consumption centers. These networks involve a large number of itinerant traders transporting small numbers of birds and live bird markets where commercial transactions occur and poultry from different places are mixed together. Importantly, the trade of infected poultry is suspected to be a driver of the propagation of major diseases like highly pathogenic avian influenza. The poultry trade has a well-established seasonal component. In Vietnam, the demand for poultry meat peaks during the lunar New Year festival period (in January-February). However, farm-level events driving the trade of poultry are poorly understood. 

We present the results of an analysis performed on the data from a longitudinal survey performed over a sample of small-scale poultry farms in southern Vietnam. Fifty-three farms were visited monthly during a 20-months period. Data related to poultry production was extracted during these monthly visits. The data included poultry population size, introduction and removal of poultry, and events affecting the flocks like mortality with specified disease symptoms. We used a mixed-effect generalized additive model enabling the prediction of the rate of harvest (sale or slaughter) of broiler chickens flocks, based on a set of independent variables. This model enabled us to specify the relationship between the age of broilers and their likelihood of being sold. In larger flocks, the slope was steeper, meaning that farmers with a larger production scale tend to sell their flocks at a faster rate. Crucially, for small flocks, the occurrence of an outbreak of disease-related mortality – in the same farm, during the same month or one month before – increased the probability of sale. This increase was even larger in case of sudden death of chickens – i.e. a suspicion of highly pathogenic avian influenza.

Share on facebook
Share on twitter
Share on linkedin

A. Tatem

The rapid ongoing changes in global connectivity are having profound effects on disease spread. Growing volumes and reach of travel and trade are enabling pathogens to move from one side of the planet to another in record time. Examples of how this rising connectivity has lead to the rapid spread of pathogens affecting people, livestock and crops are becoming prevalent, with strong evidence for the dispersal of many zoonotic viruses being human-mediated. This connectivity is not homogenous across space and time however, and an ability to capture data on human and livestock distributions, changes in these, and their movement patterns can be valuable in developing preparedness plans and models for strategic planning. Professor Tatem will present the work of WorldPop (www.worldpop.org) and collaborators in the integration of geospatial data for mapping population distributions, dynamics and connectivity, and the use of these data for obtaining epidemiological insights and planning interventions.

Share on facebook
Share on twitter
Share on linkedin

E. Valdano

INSERM, Sorbonne Université

    Trade-driven animal displacements among cattle holdings drive the likelihood, shape, and speed of outbreaks. Past works analyzed cattle networks in several countries, highlighting complex interactions between structure, function, and dynamics. A comprehensive study, linking features of cattle trade networks to their vulnerability to the spread of infectious diseases, is however still missing. Such study requires large datasets across different countries and years, to highlight global markers of cattle trade networks, as well as region-specific patterns. The main problem is data availability: cattle trade data are not public, and their access requires ad hoc agreements. I will present a collaborative platform that, using a bring code to the data approach, overcomes the strict regulations preventing data sharing, and allows an effective comparative analysis. Analyzing data from 13 European countries, we extract a set of synthetic indicators that quantifies shared features, and differences, among countries, and across years. We then show that these indicators can predict vulnerability of a specific national market to the spatial spread of a wide range of infections. Our work is a first step to building data-driven risk assessment tools that can be integrated into monitoring policies, with minimal data sharing requirements.

Share on facebook
Share on twitter
Share on linkedin

N. C. Cárdena

A. Omar, F.P. Nunes Lopes, G. Machado

1 Department of Preventive Veterinary Medicine and Animal Health, School of Veterinary Medicine and Animal Science, University of São Paulo, São Paulo, Brazil.


2 Department of Population Health and Pathobiology, College of Veterinary Medicine, North Carolina State University, Raleigh, NC, USA.
3Secretary of Agriculture, Livestock and Agribusiness of State of Rio Grande do Sul (SEAPA-RS), Porto Alegre, Brazil.

Introduction
Many infectious diseases infect multiple species and persist through a combination of within- and between-species transmission dynamics and processes. Here, we explore the epidemiological impacts of host-specific disease spread dynamics, among farms raising multiple livestock species: cattle, buffalo, pigs, sheep and goat.


Materials and methods

We reconstructed multiscale stochastic susceptible-infected network-based transmission model for within-farm dynamics and between-farm animal movements. A wide range of introduction scenarios was simulated on the empirical network. To mimic an initial stage of an introduced foreign animal disease we generated 100 runs over available 2-year network starting with 1.000 infected farms. Infection was then started randomly in farms with only swine, only cattle, only small ruminants and final scenario it started in farms with all species. The model was used to simulate control actions based on the identification of farms mode likely to be infected by its contact network.


Results

The largest epidemic had 45% infected swine farms whiten the first six months of simulation. We found that epidemic sizes were governed by which species the index was seeded to. As expected, the swine contact network was the most prone to spread disease, with a simulated prevalence of over 60% at the end of second year of simulation, followed by small ruminants’ and cattle with prevalence over 20% and 10% respectively.


Discussion

The size of epidemics initiated in cattle and small ruminants generated a higher amount of infection into other single species farm holdings. This work highlights the relevance of other than cattle farms in the between-farm transmission of possible foreign animal disease, i.e., foot and mouth disease. These results may serve as basic data in the planning of national or regional to designing risk-based targeted surveillance strategies considering a multi-species approach.

Share on facebook
Share on twitter
Share on linkedin
picture

D. Ekwem

T. Lembo, J. Enright , J. Buza, G. Shirima, R. Reeve, G. Hopcraft, T. Morrison

1 Boyd Orr Centre for Population and Ecosystem Health, Institute of Biodiversity, Animal Health and Comparative Medicine, College of Medical, Veterinary and Life Science, University of Glasgow, United Kingdom.


2 Nelson Mandela African Institution of Science and Technology, School of Life Sciences and Bioengineering, Arusha, Tanzania.

Introduction 

Livestock are essential to food security and livelihoods in sub-Saharan Africa, but suffer from poor productivity due to infectious diseases such as foot-and-mouth disease (FMD). FMD is endemic in this part of Africa, and outbreaks are frequent, driven mostly by unrestricted livestock movements. To control endemic FMD, comprehensive information on the patterns of spread through herd contacts is needed. However, data on livestock movements across the landscape, and how and where contact occurs remain limited.

 

Materials and methods 

We deployed Global Positioning System (GPS) collars on cattle in 52 different herds to understand fine-scale movements and between-herd contacts in rural areas of western Serengeti, Tanzania, representative of agropastoral systems in East Africa. We used the telemetry data to characterise the patterns of movements and identify locations of interactions between herds that suggest FMD flashpoints.  In addition, we examined patterns of contact across a range of spatiotemporal scales, relevant to different FMD transmission scenarios.

Results 

We observed that daily movement of cattle increased with herd size and rainfall. Herd contact rates were highest at large spatial and temporal scales. Furthermore, contact was greatest away from household locations, during low rainfall and close to dipping points. Generally, there were higher contacts proximal to resource areas such as grazing and water holes, but only for smaller spatiotemporal contact scales.

Discussion 

We demonstrate how widespread movements could heighten the risk of endemic FMD spread. Given that risk is directly related to contact, the probability of FMD spreading between herds could be four times higher when virus survival in the environment increases from one to up to 24 hours. Our results point at times and locations of greatest FMD transmission potential and that could be targeted through tailored control strategies, for example when rainfall levels are low, and around dipping and water points.

Share on facebook
Share on twitter
Share on linkedin

G. Guyver-Fletcher

I. Keskin, A. Bulut, K. Shea, M. Ferrari, M. Huran, X. Li, C. Jewell, E. Gorsich, M. Tildesley

1 SBIDER, The Zeeman Institute, The University of Warwick, Coventry, United Kingdom


2 Ministry of Agriculture and Forestry, Republic of Turkey
3 Penn State University, Pennsylvania, USA


4 CHICAS, Lancaster Medical School, Lancaster University, United Kingdom

Share on facebook
Share on twitter
Share on linkedin

Francis Wanyoike

F. Schirdewahn, H. H. K. Lentz, V. Colizza, A. Koher, P. Hövel,

1 Institut für Theoretische Physik, Technische Universität Berlin, Berlin, Germany
2 Institute of Epidemiology, Friedrich-Loeffler-Institut, Greifswald – Insel Riems,
Germany
3 Sorbonne Universités, UPMC Univ Paris 06, INSERM, Institut Pierre Louis
d’épidémiologie et de Santé Publique, Paris, France
4 School of Mathematical Sciences, University College Cork, Cork, Ireland
5 Veterinary Public Health Institute, University of Bern, Bern-Liebefeld, Switzerland

Introduction

Surveillance of infectious diseases in livestock is traditionally carried out at the farms, which are the typical units of epidemiological investigations and interventions. In Central and Western Europe, high-quality, long-term time series of animal transports have become available and this opens the possibility to new approaches like sentinel

surveillance. By comparing a sentinel surveillance scheme based on markets to one based on farms, the primary aim of this paper is to identify the smallest set of sentinel holdings that would reliably and timely detect emergent disease outbreaks in Swiss cattle. 

 

Materials and methods

Using a data-driven approach, we simulate the spread of infectious diseases according to the reported or available daily cattle transport data in Switzerland over a four year period. 

 

Results

Investigating the efficiency of surveillance at either market or farm level, we find that the most efficient early warning surveillance system [the smallest set of sentinels that timely and reliably detect outbreaks (small outbreaks at detection, short detection delays)] would be based on the former, rather than the latter. We show that a detection probability of 86% can be achieved by monitoring all 137 markets in the network. Additional 250 farm sentinels, selected according to their risk, need to be placed under surveillance so that the probability of first hitting one of these farm sentinels is at least as high as the probability of first hitting a market. Combining markets and 1000 farms with highest risk of infection, these two levels together will lead to a detection probability of 99%.

 

Discussion

We conclude that the design of animal surveillance systems greatly benefits from the use of the existing abundant and detailed animal transport data especially for highly dynamic cattle transport networks. Sentinel surveillance approaches can be tailored to complement existing farm risk-based and syndromic surveillance approaches.

Share on facebook
Share on twitter
Share on linkedin

Cécile Squarzoni-Diaw

Elena Arsevska , Sana Kalthoum , Pachka Hammami, Jamel Cherni †, Assia Daoudi , Mohamed Karim Laoufi , Yassir Lezaar , Kechna Rachid , Ismaila Seck , Facundo Muñoz , Renaud Lancelot, Caroline Coste

1 CIRAD, UMR ASTRE, F-34398 Montpellier, France

2 CIRAD, UMR ASTRE, F-97490 Sainte Clotilde, La Réunion, France

3 Centre national de veille zoosanitaire (CNVZ), Tunis, Tunisia

4 Ministry of Agriculture and Rural Development, Alger, Algeria

5 Office National des Sécurité Sanitaire des Produits Alimentaires (ONSSA), Rabat, Morocco

7 ASTRE, Univ Montpellier, CIRAD, INRAE, Montpellier, France

8 Food and Agricultural organization of the United Nations (FAO), Regional Office for Africa (RAF), 2 Gamel Abdul Nasser Road, PO Box GP 1628 Accra, Ghana

9Ministère de l’Élevage et des Productions Animales, Dakar, Sénégal

Introduction

We present a participative qualitative risk assessment framework to detect hotpots for risk of introduction and spread of transboundary infectious animal disease on a national scale. The framework was developed though regional training-action workshops and field activities with an involvement of national animal health services.


Material and methods

To estimate the risk of introduction, we use the epidemiological status of neighbouring countries and accessibility to major cities. In addition, we consider the highest in-degree measure of transboundary animal movements directed to a given epidemiological unit. To estimate the risk of spread we consider the highest degree and betweenness of national animal movements to detect high-risk connections between different epidemiological units. Depending on the disease, additional spatial factors such as watering points, animal density, etc. may be used. Finally, experts categorize and combine the spatial risk factors into ordinal levels of risk per epidemiological unit.


Results

We estimated the risk of introduction and spread of foot-and-mouth disease (FMD) in Tunisia as part of a series of workshops between 2015 and 2018. Out of the 2,075 Tunisian imadas, 23 were at a very high risk of FMD introduction; and 59 were at a very high risk of FMD spread. To validate the model, the results were compared to the FMD outbreaks notified by Tunisia during the 2014 FMD epizootic. Using a spatial Poisson model, we showed that the relative risk of FMD occurrence was thus 3.2 higher for imadas in the very high and high spread-risk categories than for imadas in the low and negligible spread-risk categories.


Discussion
Our results show that our framework can be a useful decision-support tool for risk-based disease surveillance and control, in particular in scarce-data environments where animal mobility has a major role of disease spread.

Share on facebook
Share on twitter
Share on linkedin

R. Aguanno

S. Von Dobschuetz1, S. Khomenko1, A. Tripodi1, W. Kalpravidh1, K. Sumption

1Food and Agriculture Organization of the United Nations, Viale delle Terme di Caracalla, Roma, Italia, 00153


2 Royal Veterinary College, University College London, 4 Royal College St, London, United Kingdom, NW1 0TU

Introduction
Studies relating to animal movement patterns and are often conducted by academic authors, with dissemination of findings limited to grey literature and scientific journals. Consequently, objectives and outcomes may not incorporate the priorities of surveillance system actors or communicate results to stakeholders. The Food and Agriculture Organisations has developed a digital and interactive medium to facilitate and empower capacity building efforts related to market profiling and animal movement under its Epidemiology Value Chain (EVC) Platform, enabling users to maintain a live, online, and dynamic tool that can store, analyse, and display a magnitude of different data.

Materials and methods
‘Open-source’ electronic collection systems (e.g.EpiCollect5) allow national veterinary services to enter value chain locations and movement patterns. Data is continuously collected through interviews, expert opinion, or retrospectively via the collation of movement permits. Various ‘plug-in’ applications allow for the visualisation of data via maps, statistics, or graphs. These are created in conjunction with national epidemiology units to ensure relevance for selecting and planning intervention options. Lastly, the application allows for dissemination of data to engage key stakeholders.

Results
Over 1000 bird markets and network connections were identified, profiled, and analysed across Viet Nam, Democratic Republic of the Congo, Ethiopia, Uganda, Rwanda, and Mozambique. Data collection in Ghana continues in conjunction with the national epidemiology unit, including expansion to livestock markets and other value chain nodes.


Discussion
The mapping of epidemiological significant locations such as markets, abattoirs, and border points along with seasonal and quantified animal movement flows can be utilised by veterinary services to plan and run prevention and control interventions. The tool can rapidly increase country capacities to identify high risk locations, i.e. those in need of urgent biosecurity improvements or those that need to be targeted through surveillance. Furthermore, data can be updated in real-time to maintain the cost efficiency and effectiveness of interventions.

Share on facebook
Share on twitter
Share on linkedin

B. Vidondo

F. Schirdewahn, H. H. K. Lentz, V. Colizza, A. Koher, P. Hövel,

1 Institut für Theoretische Physik, Technische Universität Berlin, Berlin, Germany
2 Institute of Epidemiology, Friedrich-Loeffler-Institut, Greifswald – Insel Riems,
Germany
3 Sorbonne Universités, UPMC Univ Paris 06, INSERM, Institut Pierre Louis
d’épidémiologie et de Santé Publique, Paris, France
4 School of Mathematical Sciences, University College Cork, Cork, Ireland
5 Veterinary Public Health Institute, University of Bern, Bern-Liebefeld, Switzerland

Introduction

Surveillance of infectious diseases in livestock is traditionally carried out at the farms, which are the typical units of epidemiological investigations and interventions. In Central and Western Europe, high-quality, long-term time series of animal transports have become available and this opens the possibility to new approaches like sentinel

surveillance. By comparing a sentinel surveillance scheme based on markets to one based on farms, the primary aim of this paper is to identify the smallest set of sentinel holdings that would reliably and timely detect emergent disease outbreaks in Swiss cattle. 

 

Materials and methods

Using a data-driven approach, we simulate the spread of infectious diseases according to the reported or available daily cattle transport data in Switzerland over a four year period. 

 

Results

Investigating the efficiency of surveillance at either market or farm level, we find that the most efficient early warning surveillance system [the smallest set of sentinels that timely and reliably detect outbreaks (small outbreaks at detection, short detection delays)] would be based on the former, rather than the latter. We show that a detection probability of 86% can be achieved by monitoring all 137 markets in the network. Additional 250 farm sentinels, selected according to their risk, need to be placed under surveillance so that the probability of first hitting one of these farm sentinels is at least as high as the probability of first hitting a market. Combining markets and 1000 farms with highest risk of infection, these two levels together will lead to a detection probability of 99%.

 

Discussion

We conclude that the design of animal surveillance systems greatly benefits from the use of the existing abundant and detailed animal transport data especially for highly dynamic cattle transport networks. Sentinel surveillance approaches can be tailored to complement existing farm risk-based and syndromic surveillance approaches.

Share on facebook
Share on twitter
Share on linkedin

Partners

EuFMD Open Session

Special edition

EuFMD Highlights

Contact us

EuFMD Secretariat FAO HQ, Room C-516 Viale delle Terme di Caracalla. 

Rome 00153, Italy

Click to rate this post!
[Total: 0 Average: 0]