From network analysis to risk analysis--An approach to risk-based surveillance for bovine tuberculosis in Minnesota, US

Descriptive network analysis and the influence of timescale on centrality and cohesion metrics from a system of between-herd dairy cow movements in Ontario, Canada

Previous research has demonstrated that static monthly networks of between-herd dairy cow movements in Ontario, Canada were highly fragmented, reducing potential for large-scale outbreaks. Extrapolating results from static networks can become problematic for diseases with an incubation period that exceeds the timescale of the network. The objectives of this research were to: 1) describe the networks of dairy cow movements in Ontario, and 2) describe the changes that occur among network analysis metrics when conducted at seven different timescales. Networks of dairy cow movements were created using Lactanet Canada milk recording data collected in Ontario between 2009 and 2018. Centrality and cohesion metrics were calculated after aggregating the data at seven timescales: weekly, monthly, semi-annual, annual, biennial, quinquennial, and decennial. There were 50,598 individual cows moved between Lactanet-enrolled farms, representing approximately 75% of provincially registered dairy herds. Most movements occurred over short distances (median = 39.18 km), with fewer long-range movements (maximum = 1150.80 km). The number of arcs increased marginally relative to the number of nodes with longer network timescales. Both mean out-degree, and mean clustering coefficients increased disproportionately with increasing timescale. Conversely, mean network density decreased with increasing timescale. The largest weak and strong components at the monthly timescale were small relative to the full network (267 and 4 nodes), whereas yearly networks had much higher values (2213 and 111 nodes). Higher relative connectivity in networks with longer timescales suggests pathogens with long incubation periods and animals with subclinical infection present increased potential for wide-spread disease transmission among dairy farms in Ontario. Careful consideration of disease-specific dynamics should be made when using static networks to model disease transmission among dairy cow populations.

Can Ingoing Contact Chains and other cattle movement network metrics help predict herd-level bovine tuberculosis in Irish cattle herds?

We used logistic regression to investigate whether the risk of an Irish cattle herd undergoing a bovine tuberculosis (bTB) breakdown increased with the size of the Ingoing Contact Chain (ICC) of previous herd to herd cattle movements, in a sequence up to eight moves back from the most recent, direct, movement into the herd. We further examined whether taking into account the bTB test history of each herd in the chain would improve model fit. We found that measures of cattle movements directly into the herd were risk factors for subsequent bTB restrictions, and the number of herds that animals were coming from was the most important of these. However, in contrast to a previous study in Great Britain, the ICC herd count at steps more remote than direct movements into the herd did not result in better fitting models than restricting the count to direct movements. Restricting the ICC counts to herds which had previously or would in the future test positive for bTB resulted in improved model fits, but this was not the case if only the previous test status was considered. This suggests that in many cases bTB infected animals are moving out of herds before being identified through testing, and that risk-based trading approaches should not rely solely on the previous test history of source herds as a proxy for future risk. Model fit was also improved by the inclusion of variables measuring bTB history of the herd, bTB in neighbouring herds, herd size, herd type, the movement network measures "in strength" and "betweenness", altitude, modelled badger abundance and county. Rainfall was not a good predictor. The most influential measures of bTB in nearby herds (a proxy for neighbourhood infection) were the proportion of herds with a history of bTB whose centroids were within 6 km, or whose boundaries were within 4 km, of the index herd. As well as informing national control and surveillance measures, our models can be used to identify areas where bTB rates are anomalously high, to prompt further investigation in these areas.

Assessing intrastate shipments from interstate data and expert opinion

Live animal shipments are a potential route for transmitting animal diseases between holdings and are crucial when modelling spread of infectious diseases. Yet, complete contact networks are not available in all countries, including the USA. Here, we considered a 10% sample of Interstate Certificate of Veterinary Inspections from 1 year (2009). We focused on distance dependence in contacts and investigated how different functional forms affect estimates of unobserved intrastate shipments. To further enhance our predictions, we included responses from an expert elicitation survey about the proportion of shipments moving intrastate. We used hierarchical Bayesian modelling to estimate parameters describing the kernel and effects of expert data. We considered three functional forms of spatial kernels and the inclusion or exclusion of expert data. The resulting six models were ranked by widely applicable information criterion (WAIC) and deviance information criterion (DIC) and evaluated through within- and out-of-sample validation. We showed that predictions of intrastate shipments were mildly influenced by the functional form of the spatial kernel but kernel shapes that permitted a fat tail at large distances while maintaining a plateau-shaped behaviour at short distances better were preferred. Furthermore, our study showed that expert data may not guarantee enhanced predictions when expert estimates are disparate.

Burkholderia mallei: The dynamics of networks and disease transmission

Glanders is a highly infectious zoonotic disease caused by Burkholderia mallei. The transmission of B. mallei occurs mainly by direct contact, and horses are the natural reservoir. Therefore, the identification of infection sources within horse populations and animal movements is critical to enhance disease control. Here, we analysed the dynamics of horse movements from 2014 to 2016 using network analysis in order to understand the flow of animals in two hierarchical levels, municipalities and farms. The municipality-level network was used to investigate both community clustering and the balance between the municipality's trades and the farm-level network associations between B. mallei outbreaks and the network centrality measurements, analysed by spatio-temporal generalized additive model (GAM). Causal paths were established for the dispersion of B. mallei outbreaks through the network. Our approach captured and established a direct relationship between movement of infected equines and predicted B. mallei outbreaks. The GAM model revealed that the parameters in degree and closeness centrality out were positively associated with B. mallei. In addition, we also detected 10 communities with high commerce among municipalities. The role of each municipality within the network was detailed, and significant changes in the structures of the network were detected over the course of 3 years. The results suggested the necessity to focus on structural changes of the networks over time to better control glanders disease. The identification of farms with a putative risk of B. mallei infection using the horse movement network provided a direct opportunity for disease control through active surveillance, thus minimizing economic losses and risks for human cases of B. mallei.

Model-guided suggestions for targeted surveillance based on cattle shipments in the U.S

Risk-based sampling is an essential component of livestock health surveillance because it targets resources towards sub-populations with a higher risk of infection. Risk-based surveillance in U.S. livestock is limited because the locations of high-risk herds are often unknown and data to identify high-risk herds based on shipments are often unavailable. In this study, we use a novel, data-driven network model for the shipments of cattle in the U.S. (the U.S. Animal Movement Model, USAMM) to provide surveillance suggestions for cattle imported into the U.S. from Mexico. We describe the volume and locations where cattle are imported and analyze their predicted shipment patterns to identify counties that are most likely to receive shipments of imported cattle. Our results suggest that most imported cattle are sent to relatively few counties. Surveillance at 10 counties is predicted to sample 22-34% of imported cattle while surveillance at 50 counties is predicted to sample 43%-61% of imported cattle. These findings are based on the assumption that USAMM accurately describes the shipments of imported cattle because their shipments are not tracked separately from the remainder of the U.S. herd. However, we analyze two additional datasets - Interstate Certificates of Veterinary Inspection and brand inspection data - to ensure that the characteristics of potential post-import shipments do not change on an annual scale and are not dependent on the dataset informing our analyses. Overall, these results highlight the utility of USAMM to inform targeted surveillance strategies when complete shipment information is unavailable.

Combining network analysis with epidemiological data to inform risk-based surveillance: Application to hepatitis E virus (HEV) in pigs

•

A method is proposed to explore the role of pig movements on pathogen epidemiology.

•

Pig farm centrality in the network is associated with higher HEV seroprevalence.

•

Some local areas are more at risk for HEV due to incoming pig movements.

•

Animal movements should be included in risk-based surveillance strategies.

Animal movements between farms are a major route of pathogen spread in the pig production sector. This study aimed to pair network analysis and epidemiological data in order to evaluate the impact of animal movements on pathogen prevalence in farms and assess the risk of local areas being exposed to diseases due to incoming movements. Our methodology was applied to hepatitis E virus (HEV), an emerging foodborne zoonotic agent of concern that is highly prevalent in pig farms. Firstly, the pig movement network in France (data recorded in 2013) and the results of a nation-wide seroprevalence study (data collected in 178 farms in 2009) were modelled and analysed. The link between network centrality measures of farms and HEV seroprevalence levels was explored using a generalised linear model. The in-degree and ingoing closeness of farms were found to be statistically associated with high HEV within-farm seroprevalence (p < 0.05). Secondly, the risk of a French département (i.e. French local administrative areas) being exposed to HEV was calculated by combining the distribution of farm-level HEV prevalence in source départements with the number of movements coming from those same départements. By doing so, the risk of exposure for départements was mapped, highlighting differences between geographical patterns of HEV prevalence and the risk of exposure to HEV. These results suggest that not only highly prevalent areas but also those having at-risk movements from infected areas should be monitored. Pathogen management and surveillance options in the pig production sector should therefore take animal movements into consideration, paving the way for the development of targeted and risk-based disease surveillance strategies.

Evaluating empirical contact networks as potential transmission pathways for infectious diseases

Networks are often used to incorporate heterogeneity in contact patterns in mathematical models of pathogen spread. However, few tools exist to evaluate whether potential transmission pathways in a population are adequately represented by an observed contact network. Here, we describe a novel permutation-based approach, the network k-test, to determine whether the pattern of cases within the observed contact network are likely to have resulted from transmission processes in the network, indicating that the network represents potential transmission pathways between nodes. Using simulated data of pathogen spread, we compare the power of this approach to other commonly used analytical methods. We test the robustness of this technique across common sampling constraints, including undetected cases, unobserved individuals and missing interaction data. We also demonstrate the application of this technique in two case studies of livestock and wildlife networks. We show that the power of the k-test to correctly identify the epidemiologic relevance of contact networks is substantially greater than other methods, even when 50% of contact or case data are missing. We further demonstrate that the impact of missing data on network analysis depends on the structure of the network and the type of missing data.

Animal movement network analysis as a tool to map farms serving as contamination source in cattle cysticercosis

ABSTRACT: Bovine cysticercosis is a problem distributed worldwide that result in economic losses mainly due to the condemnation of infected carcasses. One of the difficulties in applying control measures is the identification of the source of infection, especially because cattle are typically acquired from multiple farms. Here, we tested the utility of an animal movement network constructed with data from a farm that acquires cattle from several other different farms to map the major contributors of cysticercosis propagation. Additionally, based on the results of the network analysis, we deployed a sanitary management and drug treatment scheme to decrease cysticercosis’ occurrence in the farm. Six farms that had commercial trades were identified by the animal movement network and characterized as the main contributors to the occurrence of cysticercosis in the studied farm. The identification of farms with a putative risk of Taenia saginata infection using the animal movement network along with the proper sanitary management and drug treatment resulted in a gradual decrease in cysticercosis prevalence, from 25% in 2010 to 3.7% in 2011 and 1.8% in 2012. These results suggest that the animal movement network can contribute towards controlling bovine cysticercosis, thus minimizing economic losses and preventing human taeniasis.

Epidemiological investigation of bovine tuberculosis outbreaks in Uruguay (2011-2013)

Bovine tuberculosis (bTB) is a chronic disease of cattle caused by infection with the Mycobacterium bovis. While bTB prevalence in Uruguay has been low (<11 outbreaks/year) for the past 50 years as a consequence of a national control program, annual incidence increased in 2011 through 2013-15, 26 and 16 infected herds each year, raising concerns from livestock stakeholders and the government. The goal of this study was to assess the spatial dynamics of bTB in Uruguay from 2011 to 2013 and the association between bTB and potential demographic and movement risk factors at the herd level using data provided by the Uruguayan Ministry of Livestock, Agriculture, and Fisheries. Clustering of incident outbreaks was assessed using the Cuzick-Edwards' test and the Bernoulli model of the spatial scan statistic, and a conditional multivariable logistic regression model was used to assess risk factors associated with bTB in a subset of Uruguayan dairy farms. Significant (P<0.05) global clustering was detected in 2012, while high-risk local clusters were detected in southwestern (2011, 2012, 2013), northwestern (2012), and southeastern (2012) Uruguay. Increased risk of bTB in different regions of Uruguay suggests a potential role of animal movements in disease dissemination. Larger herds, higher numbers of animals purchased, and incoming steers to the farm were associated with increased odds of breaking with bTB, in agreement with previous studies but also suggesting other additional sources of risk. These results will contribute to enhanced effectiveness of bTB control programs in Uruguay with the ultimate objective of preventing or mitigating the impact of the disease in the human and animal populations of the country.

Role of Cattle Movements in Bovine Tuberculosis Spread in France between 2005 and 2014

Live animal movements are a major transmission route for the spread of infectious agents such as Mycobacterium bovis, the main agent of bovine Tuberculosis (bTB). France became officially bTB-free in 2001, but M. bovis is still circulating in the cattle population, with about a hundred of outbreaks per year, most located in a few geographic areas. The aim of this study was to analyse the role of cattle movements in bTB spread in France between 2005 and 2014, using social network analysis and logistic regression models. At a global scale, the trade network was studied to assess the association between several centrality measures and bTB infection though a case-control analysis. The bTB infection status was associated with a higher in-degree (odds-ratio [OR] = 2.4 [1.1–5.4]) and with a higher ingoing contact chain (OR = 2.2 [1.0–4.7]). At a more local scale, a second case-control analysis was conducted to estimate the relative importance of cattle movements and spatial neighbourhood. Only direct purchase from infected herds was shown to be associated with bTB infection (OR = 2.9 [1.7–5.2]), spatial proximity to infected herds being the predominant risk factor, with decreasing ORs when distance increases. Indeed, the population attributable fraction was 12% [5%–18%] for cattle movements and 73% [68%–78%] for spatial neighbourhood. Based on these results, networks of potential effective contacts between herds were built and analysed for the three major spoligotypes reported in France. In these networks, the links representing cattle movements were associated with higher edge betweenness than those representing the spatial proximity between infected herds. They were often links connecting distinct communities and sometimes distinct geographical areas. Therefore, although their role was quantitatively lower than the one of spatial neighbourhood, cattle movements appear to have been essential in the French bTB dynamics between 2005 and 2014.

Network, cluster and risk factor analyses for porcine reproductive and respiratory syndrome using data from swine sites participating in a disease control program

The objectives of this study were to describe networks of Ontario swine sites and their service providers (including trucking, feed, semen, gilt and boar companies); to categorize swine sites into clusters based on site-level centrality measures, and to investigate risk factors for porcine reproductive and respiratory syndrome (PRRS) using information gathered from the above-mentioned analyses. All 816 sites included in the current study were enrolled in the PRRS area regional control and elimination projects in Ontario. Demographics, biosecurity and network data were collected using a standardized questionnaire and PRRS status was determined on the basis of available diagnostic tests and assessment by site veterinarians. Two-mode networks were transformed into one-mode dichotomized networks. Cluster and risk factor analyses were conducted separately for breeding and growing pig sites. In addition to the clusters obtained from cluster analyses, other explanatory variables of interest included: production type, type of animal flow, use of a shower facility, and number of neighboring swine sites within 3km. Unadjusted univariable analyses were followed by two types of adjusted models (adjusted for production systems): a generalizing estimation equation model (GEE) and a generalized linear mixed model (GLMM). Results showed that the gilt network was the most fragmented network, followed by the boar and truck networks. Considering all networks simultaneously, approximately 94% of all swine sites were indirectly connected. Unadjusted risk factor analyses showed significant associations between almost all predictors of interest and PRRS positivity, but these disappeared once production system was taken into consideration. Finally, the vast majority of the variation on PRRS status was explained by production system according to GLMM, which shows the highly correlated nature of the data, and raises the point that interventions at this level could potentially have high impact in PRRS status change and/or maintenance.

Evaluation of farm-level parameters derived from animal movements for use in risk-based surveillance programmes of cattle in Switzerland

Background

This study focused on the descriptive analysis of cattle movements and farm-level parameters derived from cattle movements, which are considered to be generically suitable for risk-based surveillance systems in Switzerland for diseases where animal movements constitute an important risk pathway.

Methods

A framework was developed to select farms for surveillance based on a risk score summarizing 5 parameters. The proposed framework was validated using data from the bovine viral diarrhoea (BVD) surveillance programme in 2013.

Results

A cumulative score was calculated per farm, including the following parameters; the maximum monthly ingoing contact chain (in 2012), the average number of animals per incoming movement, use of mixed alpine pastures and the number of weeks in 2012 a farm had movements registered. The final score for the farm depended on the distribution of the parameters. Different cut offs; 50, 90, 95 and 99 %, were explored. The final scores ranged between 0 and 5. Validation of the scores against results from the BVD surveillance programme 2013 gave promising results for setting the cut off for each of the five selected farm level criteria at the 50th percentile. Restricting testing to farms with a score ≥ 2 would have resulted in the same number of detected BVD positive farms as testing all farms, i.e., the outcome of the 2013 surveillance programme could have been reached with a smaller survey.

Conclusions

The seasonality and time dependency of the activity of single farms in the networks requires a careful assessment of the actual time period included to determine farm level criteria. However, selecting farms in the sample for risk-based surveillance can be optimized with the proposed scoring system. The system was validated using data from the BVD eradication program. The proposed method is a promising framework for the selection of farms according to the risk of infection based on animal movements.

Infectious disease transmission and contact networks in wildlife and livestock

The use of social and contact networks to answer basic and applied questions about infectious disease transmission in wildlife and livestock is receiving increased attention. Through social network analysis, we understand that wild animal and livestock populations, including farmed fish and poultry, often have a heterogeneous contact structure owing to social structure or trade networks. Network modelling is a flexible tool used to capture the heterogeneous contacts of a population in order to test hypotheses about the mechanisms of disease transmission, simulate and predict disease spread, and test disease control strategies. This review highlights how to use animal contact data, including social networks, for network modelling, and emphasizes that researchers should have a pathogen of interest in mind before collecting or using contact data. This paper describes the rising popularity of network approaches for understanding transmission dynamics in wild animal and livestock populations; discusses the common mismatch between contact networks as measured in animal behaviour and relevant parasites to match those networks; and highlights knowledge gaps in how to collect and analyse contact data. Opportunities for the future include increased attention to experiments, pathogen genetic markers and novel computational tools.