Event Abstract

Back to Event

Combining multicriteria decision analysis and network-based model to assess the vulnerability of commercial Cuban poultry to avian influenza viruses

Carla Ippoliti1*, Luca Candeloro1, Lara Savini1, Annamaria Conte1, Marius Gilbert2, Joel Ayala3, Osvaldo Fonseca Rodríguez3, 4, Damarys Montano3, Amarilys Angery5 and Pastor Alfonso Zamora3
  • 1 Experimental Zooprophylactic Institute of Abruzzo and Molise G. Caporale, Italy
  • 2 Spatial Epidemiology Lab, Free University of Brussels, Belgium
  • 3 National Center for Animal and Plant Health (CENSA), Cuba
  • 4 Department of Epidemiology and Global Health, Umeå University, Sweden
  • 5 Ministry of Agriculture (Cuba), Cuba

Background The last two decades have witnessed a dramatic change in Avian influenza (AI) epidemiology in terms of affected countries as well as in a growing list of AI virus (AIV) strains that sometimes severely infect humans. For countries in which the disease has not occurred, it is difficult to assess the potential consequences of an outbreak and, as a result, it is hard to develop cost-effective prevention/eradication strategies (Dubé et al., 2009, Martínez-López et al., 2009). In absence-data situations, knowledge-based methods are the only applicable ones to predict suitability for disease occurrence; they rely on expertise and literature, and GIS-based multicriteria decision analysis (MCDA) is considered a valuable method (Stevens & Pfeiffer 2011). Moreover, in case of agent introduction, the potential epidemic size highly depends on livestock trade, i.e. on the animal movements between farms (Dubè et al, 2011; Martínez-López et al, 2009). Therefore, the integration of spatial suitability analysis for AI occurrence and a Susceptible-Infected-Susceptible (SIS) network-based model to infer AIV diffusion may allow implementing differential surveillance, prevention and control strategies (Maurella et al, 2019). Cuba is the largest (110,886 km2) and most Westerly Island of the Caribbean, accounting for about 50% of the region’s land area. The importance of Cuban wetlands for wintering waterbirds is enhanced by the fact that the island is in the middle of two important bird migration corridors, the Atlantic and the Mississippi flyways. Both corridors have been seen as important for AI diffusion in North America (Fries et al., 2015, Belkhiria et al., 2016). Moreover, AI suitability through these corridors predicted most of the outbreak areas during the 2014–2015 epidemic in the US (Belkhiria et al., 2016). The prevalence of influenza virus in general, as well as the specific distribution of subtypes, vary between different surveillance studies depending on species, time, and location (Olsen et al., 2006). However, it is reasonable to assume that from a migratory wild bird standpoint AIV prevalence in Cuba would be similar to the existing in the Southeastern US. Cuba is not only an important wintering area for migrants birds from North America, but also a very important staging area for birds that are migrating further south (Mugica et al., 2006). Suitability for aquatic birds in Cuba is extended by natural and man-made freshwater bodies, including rice-growing areas, where these birds foraging (Acosta et al., 2010). The Cuban suitability for wild birds, as primary reservoirs for avian influenza viruses (AIVs), may expose poultry to contract the infection. The poultry production in Cuba is an important component of livestock economy and food security. The Cuban poultry stock stand for over 32,3 million-head (including hens, ducks, turkeys, quails, among others) with their own breeders (AEC 2017). Commercial poultry is mainly towards egg production based on seven million of hens with an equivalent amount of pullets as replacements. The backyard type represents near 56% of total poultry stock but it is mainly raised for owner’s self-consumption. An AI risk-based surveillance strategy was already developed in Western Cuba considering proximity of poultry premises to waterbird settlements as main risk factor (Ferrer et al., 2014). Notwithstanding, it is expected that other factors in combination with waterfowl abundance, may create areas suitable for incursion and spread of AIVs in poultry population in Cuba. The objective of this study was to evaluate and quantify the probability of introduction of AIVs in the Cuban poultry industry. Being avian influenza (AI) exotic for Cuba, this study was carried out through the two following analysis steps: 1. elicitation and collection of expert opinions to rank the main risk factors associated with the onset of the disease and contributing to the probability of introduction of AIV derived through the MCDA; 2. integration of the previous results with the contribution of animal shipments between poultry farms, calculating the risk of AIV spread through a temporal network-based SIS model. Data and methods The MCDA was implemented through the following steps: 1. Identification of risk factors from literature 2. Definition of the relationship between each factor and suitability through an international expert elicitation 3. Data collection from available sources for the layers (roads, human population, waterbodies, ocean ports and marines, rice-growing areas and natural wetlands, commercial poultry density, backyard poultry density, commercial duck density, abundance of wild waterfowls) and standardization of the factors 4. Combination of the factors to produce a final suitability map using a Weighted Linear Combination. The list of layers, data sources, spatial resolutions and data manipulation methods are reported in table 1. Once the raster covering the Cuban Archipelago at 1 km spatial resolution was created, a SIS network-based model was elaborated on poultry trade data. Data on poultry shipments collected for the years 2015-2017 was provided by Technological Poultry Division (DTA) of the Cuban Ministry of Agriculture. Each recorded data includes origin and destination farm of the movement, the movement date, number of transported animals and type of the farm (S: Starter, D: development and L: laying farms). Data has been spatially aggregated at pixel level (1 km spatial resolution), thus implying the existence of pixels (nodes) that involve more types of farm. The SIS network-based model assumptions are: 1. The AI, being exotic for Cuba, may principally occur because of waterfowls (main AIV reservoir) that transit and settle in Cuba during migration periods. AIV introduction into one of the poultry farm may occur during the maximal abundance of waterfowl, in Spring (01 February - 30 April) and Autumn (01 October - 31 December) periods. 2. For each pixel, the probability of being infected is pi =1-(1-p)^0.5, where p is the risk of infection introduction derived from MCDA. Infected status for a pixel is associated with the corresponding day in which the event occurs, randomly sampling the day of the infection from a uniform distribution. 3. We used 21 days AI incubation period according to the OIE sanitary code. After such a period, an infected pixel turns again into a susceptible state. In each simulation run and for each pixel, two values of infection status are sampled from a binomial distribution with probability pi. Network movements involving infected pixels drive the spread according to the contact timing. A pixel can get infected either through pi or through network movements. If both routes occur and the corresponding incubation periods overlap, the first date of infection is taken into account. At the end of the simulation runs, a posterior probability of being infected is summarized as the fraction of times the pixel resulted infected (p*). Geographical analysis was performed in ArcMap 10.5 ESRI®, SNA analysis in R software version 3.3.2 (https://www.r-project.org/) using the package igraph (Csardi, 2006). Results Figure 1 reports the map showing the vulnerability of Cuba to AIV introduction (p) at pixel level of 1 km spatial resolution, considering the list of factors and weights indicated by the experts and reported in Table 1. In the map, the zones with the highest vulnerability are orange-red colored, located where all the “risky” factors co-occurs, while the green color indicates the lowest level of vulnerability, i.e. few or none of “risky” factor are present. Poultry trade is mainly characterized by the commercial chain in which, the trade pattern originating from Starter and ending to Layer, passing through Development (S-> D-> L), accounts for the 61% of all trade connections (Table 2). A consistent part (around 35%) of the movements regards the direct connections from S to L. Network differences among the years 2015, 2016 and 2017 resulted negligible. The simulations of the SIS model have been performed for several years (from 2015 to 2017) running 10,000 iterations. Figure 2 reports the scatter plot of the simulated probability of being infected (p*) versus p for pixels including nodes of type S, D and L using color coding for the years. As expected for the characteristic production chain of the country, the simulated probability always increases the p value for farms of type D and L. Farm type S shows no variation of risk due to the few incoming animal movements (they are egg-production farms). The geographical distribution of poultry farms on Cuba Archipelago is shown in Figure 3, reporting the farm types: S (triangle), D (circles) and L (squares). The color varies according to diff=p*-p, i.e. the amount of change of the risk due to space and time occurrence of poultry movements in the pixel where the farm is located. Figure 4 shows the Pearson correlation between diff and network centrality measures. Low correlation values strengthen the importance of SIS model beyond the basic centrality measures, among which InStrength and PageRank showed the highest values. Conclusions This work presents a methodology to integrate spatial suitability analysis for AI occurrence and a SIS network-based model to infer AIV diffusion in absence of the disease. Such approach will be useful for the implementation of targeted surveillance, prevention, and control strategies at specific periods and locations, which will benefit in the optimization of human and financial resources.

Figure 1
Image
Figure 2
Image
Figure 3
Image
Figure 4
Image
Figure 5
Image
Figure 6
Image
Figure 7
Image

Acknowledgements

We greatly appreciate the cooperation of Martin Acosta and Lourdes Mugica for providing information on the abundance of Anseriformes. We thank Technological Poultry Division (DTA) of the Cuban Ministry of Agricultural for providing data on national poultry shipments, as well National Veterinary Service for providing location (geographical coordinates) and data from each poultry commercial farms registered.

References

1. Acosta M, Mugica L, Blanco D, López-Lanùs B, Antunes R. Doodnath L, and Hurtado J. Birds of Rice Fields in the Americas, Waterbirds, 33(1):105-122. 2010. 2. AEC 2017. Anuario Estadístico de Cuba CAPÍTULO 9: AGRICULTURA, GANADERÍA, SILVICULTURA Y PESCA. Available at: http://www.one.cu/aec2016/09%20Agricultura%20Ganaderia%20Silvicultura%20Pesca.pdf 3. Belkhiria, J. et al. Application of Species Distribution Modeling for Avian Influenza surveillance in the United States considering the North America Migratory Flyways. Sci. Rep. 6, 33161; doi: 10.1038/srep33161 (2016). 4. Buttner K, Krieter J, Traulsen A, Traulsen I (2013) Efficient Interruption of Infection Chains by Targeted Removal of Central Holdings in an Animal Trade Network. PLoS ONE 8(9): e74292. doi:10.1371/journal.pone.0074292 5. Butts CT, Hunter D, Handcock M, Bender-deMoll S, Horner J, Wang L. network: Classes for Relational Data. 2019 [cited 2019 May 30]. Available from: https://CRAN.R-project.org/package=network 6. Dubé C, Ribble C, Kelton D, McNab B. Introduction to network analysis and its implications for animal disease modelling. Rev - Off Int Epizoot. 2011 Aug;30(2):425–36. 7. Dubé C., Ribble C., Kelton D. and McNab B. A Review of Network Analysis Terminology and its Application to Foot-and-Mouth Disease Modelling and Policy Development. Transboundary and Emerging Diseases. 56 (2009) 73–85. doi:10.1111/j.1865-1682.2008.01064.x 8. Ferrer E., Alfonso P., Ippoliti C., Abeledo M., Calistri P., Blanco P., Conte A., Sánchez B., Fonseca O., Percedo M., Pérez A., Fernández O., Giovannini A. 2014. Development of an active risk-based surveillance strategy for avian influenza in Cuba. Prev Vet Med DOI: 10.1016/j.prevetmed.2014.05.012 9. Fries AC, Nolting JM, Bowman AS, Lin X, Halpin RA, Wester E, Fedorova N, Stockwell TB, Das SR, Dugan VG, Wentworth DE, Gibbs HL, Slemons RD. 2015. Spread and persistence of influenza A viruses in waterfowl hosts in the North American Mississippi Migratory Flyway. J Virol 89:5371–5381. doi:10.1128/JVI.03249-14. 10. Martínez-López B, Perez AM, Sánchez-Vizcaíno JM. Social network analysis. Review of general concepts and use in preventive veterinary medicine. Transbound Emerg Dis. 2009 May;56(4):109–20 11. Maurella C, Mastrantonio G, Bertolini S, Crescio MI, Ingravalle F, Adkin A, et al. Social network analysis and risk assessment: An example of introducing an exotic animal disease in Italy. Microbial Risk Analysis [Internet]. 2019 Apr 3 [cited 2019 May 30]; Available from: http://www.sciencedirect.com/science/article/pii/S2352352218300586 12. Mugica, L., Acosta, M., Denis, D., Jiménez, A., Rodríguez, A. & Ruiz, X. 2006. Rice culture in Cuba as an important wintering site for migrant waterbirds from North America. Waterbirds around the world. Eds. G.C. Boere, C.A. Galbraith & D.A. Stroud. The Stationery Office, Edinburgh, UK. pp. 172-176Olsen, B., Munster, V. J., Wallensten, A., Waldenström, J., Osterhaus, A. D., & Fouchier, R. A. (2006). Global patterns of influenza A virus in wild birds. science, 312(5772), 384-388. https://doi.org/10.1126/science.1122438 13. Stevens K.B. & Pfeiffer D.U. Spatial modelling of disease using data- and knowledge-driven approaches. Spatial and Spatio-temporal Epidemiology 2 (2011) 125–133 14. Loth L, Gilbert M, Osmani MG, Kalam AM, Xiao X. Risk factors and clusters of Highly Pathogenic Avian Influenza H5N1 outbreaks in Bangladesh. Preventive Veterinary Medicine. 2010 Aug 1;96(1):104–13 15. Csardi G, Nepusz T: The igraph software package for complex network research, InterJournal, Complex Systems 1695. 2006. http://igraph.org

Keywords: Avian influenza (AI), Cuba, Multicriteria decision analysis (MCDA), SIS model, Network analysis

Conference: GeoVet 2019. Novel spatio-temporal approaches in the era of Big Data, Davis, United States, 8 Oct - 10 Oct, 2019.

Presentation Type: Regular oral presentation

Topic: Spatial-explicit or spatio-temporal network analysis

Citation: Ippoliti C, Candeloro L, Savini L, Conte A, Gilbert M, Ayala J, Fonseca Rodríguez O, Montano D, Angery A and Alfonso Zamora P (2019). Combining multicriteria decision analysis and network-based model to assess the vulnerability of commercial Cuban poultry to avian influenza viruses. Front. Vet. Sci. Conference Abstract: GeoVet 2019. Novel spatio-temporal approaches in the era of Big Data. doi: 10.3389/conf.fvets.2019.05.00110

Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.

The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.

Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.

For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.

Received: 10 Jun 2019; Published Online: 27 Sep 2019.

* Correspondence: Mx. Carla Ippoliti, Experimental Zooprophylactic Institute of Abruzzo and Molise G. Caporale, Teramo, Abruzzo, Italy, c.ippoliti@izs.it