Use of Network Analysis and Spread Models to Target Control Actions for Bovine Tuberculosis in a State from Brazil

Abstract

Livestock movements create complex dynamic interactions among premises that can be represented, interpreted, and used for epidemiological purposes. These movements are a very important part of the production chain but may also contribute to the spread of infectious diseases through the transfer of infected animals over large distances. Social network analysis (SNA) can be used to characterize cattle trade patterns and to identify highly connected premises that may act as hubs in the movement network, which could be subjected to targeted control measures in order to reduce the transmission of communicable diseases such as bovine tuberculosis (TB). Here, we analyzed data on cattle movement and slaughterhouse surveillance for detection of TB-like lesions (TLL) over the 2016–2018 period in the state of Rio Grande do Sul (RS) in Brazil with the following aims: (i) to characterize cattle trade describing the static full, yearly, and monthly snapshots of the network contact trade, (ii) to identify clusters in the space and contact networks of premises from which animals with TLL originated, and (iii) to evaluate the potential of targeted control actions to decrease TB spread in the cattle population of RS using a stochastic metapopulation disease transmission model that simulated within-farm and between-farm disease spread. We found heterogeneous densities of premises and animals in the study area. The analysis of the contact network revealed a highly connected (~94%) trade network, with strong temporal trends, especially for May and November. The TLL cases were significantly clustered in space and in the contact network, suggesting the potential for both local (e.g., fence-to-fence) and movement-mediated TB transmission. According to the disease spread model, removing the top 7% connected farms based on degree and betweenness could reduce the total number of infected farms over three years by >50%. In conclusion, the characterization of the cattle network suggests that highly connected farms may play a role in TB dissemination, although being close to infected farms was also identified as a risk factor for having animals with TLL. Surveillance and control actions based on degree and betweenness could be useful to break the transmission cycle between premises in RS.

Multiple species animal movements: network properties, disease dynamics and the impact of targeted control actions

“…Additionally, the degree distribution results could help highlight super-spreader farms in all host populations through the identification of farms with several commercial partners that make a major contribution to disease spread [ 41 ]. For example, the swine network was highlighted as the network with the largest degree, which has been described previously in the same region [ 2 , 21 ]. Finally, the analysis of the small ruminant networks showed a very disconnected network with a marked seasonality at the end of each year but with fewer monthly edges and nodes than swine and cattle farms, despite having more registered farms than the swine production in the full network.…”

“…Therefore, we argue for the need to consider temporal models, such as our contact chain analysis, when developing network-based target approaches, because it considers the temporal order of the connections and avoids over-connectivity of the static networks [ 20 , 45 ]. Furthermore, network analysis approaches have been widely used recently; the great majority centered on describing the contact networks of a single species, here we cite just some relevant examples [ 2 , 18 , 32 , 39 , 40 , 46 – 48 ], however, only a fraction of studies have explicitly considered the interaction of possible contacts among more than one species while constructing their transmission networks [ 7 – 9 , 11 – 13 , 15 , 16 , 49 ]. While single host networks are informative, they are likely to underestimate the epidemic propagation of pathogens capable of infecting multiple species [ 26 , 50 – 53 ].…”

“…Historically, these epidemics have been associated with disease transmission among multiple livestock hosts, i.e., the 2001 foot-and-mouth disease (FMD) outbreak in the United Kingdom [ 1 ]. To better understand the main drivers of such large-scale outbreaks, several studies have used social network analysis (SNA) and mathematical models to characterize animal movement patterns, often with a focus on individual food animal species such as cattle, [ 2 – 5 ] swine or small ruminants [ 6 ]. Only a few studies have developed mathematical models or even described contact networks over animal movement data of multiple species at the same time [ 7 – 16 ].…”

“…In previous literature, social network analysis has been used to shed light on between-farm disease spread processes, including the calculation of expected epidemic size where a wide range of network features derived from both static and monthly snapshots of network representation have been implemented [ 2 , 17 , 18 ]. The use of such approaches is very useful to describe the topology and temporal variation in the networks, however, it can lead to overestimation of the connectivity of the animal trade.…”

Cattle Shipments and Disease Spread Modeling

“…In the E.U., livestock animal movements are (as mentioned above) recorded in a computerized database. Other countries also require reporting of livestock animal movements, such as Argentina (Aznar et al 2011) and Brazil (Cardenas et al 2021) whereas such requirements may not exist for all countries in Africa (Teissier 2021). The latter is also true for the U.S. where movement data are not readily available.…”