Disease prevention versus data privacy: using landcover maps to inform spatial epidemic models

A scoping review of foot-and-mouth disease risk, based on spatial and spatio-temporal analysis of outbreaks in endemic settings

Foot-and-mouth disease (FMD) is one of the most important transboundary animal diseases affecting livestock and wildlife species worldwide. Sustained viral circulation, as evidenced by serological surveys and the recurrence of outbreaks, suggests endemic transmission cycles in some parts of Africa, Asia and the Middle East. This is the result of a complex process in which multiple serotypes, multi-host interactions and numerous socio-epidemiological factors converge to facilitate disease introduction, survival and spread. Spatial and spatio-temporal analyses have been increasingly used to explore the burden of the disease by identifying high-risk areas, analysing temporal trends and exploring the factors that contribute to the outbreaks. We systematically retrieved spatial and spatial-temporal studies on FMD outbreaks to summarize variations on their methodological approaches and identify the epidemiological factors associated with the outbreaks in endemic contexts. Fifty-one studies were included in the final review. A high proportion of papers described and visualized the outbreaks (72.5%) and 49.0% used one or more approaches to study their spatial, temporal and spatio-temporal aggregation. The epidemiological aspects commonly linked to FMD risk are broadly categorizable into themes such as (a) animal demographics and interactions, (b) spatial accessibility, (c) trade, (d) socio-economic and (e) environmental factors. The consistency of these themes across studies underlines the different pathways in which the virus is sustained in endemic areas, with the potential to exploit them to design tailored evidence based-control programmes for the local needs. There was limited data linking the socio-economics of communities and modelled FMD outbreaks, leaving a gap in the current knowledge. A thorough analysis of FMD outbreaks requires a systemic view as multiple epidemiological factors contribute to viral circulation and may improve the accuracy of disease mapping. Future studies should explore the links between socio-economic and epidemiological factors as a foundation for translating the identified opportunities into interventions to improve the outcomes of FMD surveillance and control initiatives in endemic contexts.

Spatial aggregation choice in the era of digital and administrative surveillance data

Traditional disease surveillance is increasingly being complemented by data from non-traditional sources like medical claims, electronic health records, and participatory syndromic data platforms. As non-traditional data are often collected at the individual-level and are convenience samples from a population, choices must be made on the aggregation of these data for epidemiological inference. Our study seeks to understand the influence of spatial aggregation choice on our understanding of disease spread with a case study of influenza-like illness in the United States. Using U.S. medical claims data from 2002 to 2009, we examined the epidemic source location, onset and peak season timing, and epidemic duration of influenza seasons for data aggregated to the county and state scales. We also compared spatial autocorrelation and tested the relative magnitude of spatial aggregation differences between onset and peak measures of disease burden. We found discrepancies in the inferred epidemic source locations and estimated influenza season onsets and peaks when comparing county and state-level data. Spatial autocorrelation was detected across more expansive geographic ranges during the peak season as compared to the early flu season, and there were greater spatial aggregation differences in early season measures as well. Epidemiological inferences are more sensitive to spatial scale early on during U.S. influenza seasons, when there is greater heterogeneity in timing, intensity, and geographic spread of the epidemics. Users of non-traditional disease surveillance should carefully consider how to extract accurate disease signals from finer-scaled data for early use in disease outbreaks.

Modelling the domestic poultry population in the United States: A novel approach leveraging remote sensing and synthetic data methods

Comprehensive and spatially accurate poultry population demographic data do not currently exist in the United States; however, these data are critically needed to adequately prepare for, and efficiently respond to and manage disease outbreaks. In response to absence of these data, this study developed a national-level poultry population dataset by using a novel combination of remote sensing and probabilistic modelling methodologies. The Farm Location and Agricultural Production Simulator (FLAPS) (Burdett et al., 2015) was used to provide baseline national-scale data depicting the simulated locations and populations of individual poultry operations. Remote sensing methods (identification using aerial imagery) were used to identify actual locations of buildings having the characteristic size and shape of commercial poultry barns. This approach was applied to 594 U.S. counties with > 100,000 birds in 34 states based on the 2012 U.S. Department of Agriculture (USDA), National Agricultural Statistics Service (NASS), Census of Agriculture (CoA). The two methods were integrated in a hybrid approach to develop an automated machine learning process to locate commercial poultry operations and predict the number and type of poultry for each operation across the coterminous United States. Validation illustrated that the hybrid model had higher locational accuracy and more realistic distribution and density patterns when compared to purely simulated data. The resulting national poultry population dataset has significant potential for application in animal disease spread modelling, surveillance, emergency planning and response, economics, and other fields, providing a versatile asset for further agricultural research.

Challenges to the Application of Spatially Explicit Stochastic Simulation Models for Foot-and-Mouth Disease Control in Endemic Settings: A Systematic Review

Simulation modeling has become common for estimating the spread of highly contagious animal diseases. Several models have been developed to mimic the spread of foot-and-mouth disease (FMD) in specific regions or countries, conduct risk assessment, analyze outbreaks using historical data or hypothetical scenarios, assist in policy decisions during epidemics, formulate preparedness plans, and evaluate economic impacts. Majority of the available FMD simulation models were designed for and applied in disease-free countries, while there has been limited use of such models in FMD endemic countries. This paper's objective was to report the findings from a study conducted to review the existing published original research literature on spatially explicit stochastic simulation (SESS) models of FMD spread, focusing on assessing these models for their potential use in endemic settings. The goal was to identify the specific components of endemic FMD needed to adapt these SESS models for their potential application in FMD endemic settings. This systematic review followed the PRISMA guidelines, and three databases were searched, which resulted in 1176 citations. Eighty citations finally met the inclusion criteria and were included in the qualitative synthesis, identifying nine unique SESS models. These SESS models were assessed for their potential application in endemic settings. The assessed SESS models can be adapted for use in FMD endemic countries by modifying the underlying code to include multiple cocirculating serotypes, routine prophylactic vaccination (RPV), and livestock population dynamics to more realistically mimic the endemic characteristics of FMD. The application of SESS models in endemic settings will help evaluate strategies for FMD control, which will improve livestock health, provide economic gains for producers, help alleviate poverty and hunger, and will complement efforts to achieve the Sustainable Development Goals.

Poultry farm distribution models developed along a gradient of intensification

Efficient planning of measures limiting epidemic spread requires information on farm locations and sizes (number of animals per farm). However, such data are rarely available. The intensification process which is operating in most low- and middle-income countries (LMICs), comes together with a spatial clustering of farms, a characteristic epidemiological models are sensitive to. We developed farm distribution models predicting both the location and the number of animals per farm, while accounting for the spatial clustering of farms in data-poor countries, using poultry production as an example. We selected four countries, Nigeria, Thailand, Argentina and Belgium, along a gradient of intensification expressed by the per capita Gross Domestic Product (GDP). First, we investigated the distribution of chicken farms along the spectrum of intensification. Second, we built farm distribution models (FDM) based on censuses of commercial farms of each of the four countries, using point pattern and random forest models. As an external validation, we predicted farm locations and sizes in Bangladesh. The number of chicken per farm increased gradually in line with the gradient of GDP per capita in the following order: Nigeria, Thailand, Argentina and Belgium. Interestingly, we did not find such a gradient for farm clustering. Our modelling procedure could only partly reproduce the observed datasets in each of the four sample countries in internal validation. However, in the external validation, the clustering of farms could not be reproduced and the spatial predictors poorly explained the number and location of farms and farm sizes in Bangladesh. Further improvements of the methodology should explore other covariates of the intensity of farms and farm sizes, as well as improvements of the methodology. Structural transformation, economic development and environmental conditions are essential characteristics to consider for an extrapolation of our FDM procedure, as generalisation appeared challenging. We believe the FDM procedure could ultimately be used as a predictive tool in data-poor countries.

Challenges and opportunities for using national animal datasets to support foot-and-mouth disease control

National level databases of animal numbers, locations and movements provide the essential foundations for disease preparedness, outbreak investigations and control activities. These activities are particularly important for managing and mitigating the risks of high-impact transboundary animal disease outbreaks such as foot-and-mouth disease (FMD), which can significantly affect international trade access and domestic food security. In countries where livestock production systems are heavily subsidized by the government, producers are often required to provide detailed animal movement and demographic data as a condition of business. In the remaining countries, it can be difficult to maintain these types of databases and impossible to estimate the extent of missing or inaccurate information due to the absence of gold standard datasets for comparison. Consequently, competent authorities are often required to make decisions about disease preparedness and control based on available data, which may result in suboptimal outcomes for their livestock industries. It is important to understand the limitations of poor data quality as well as the range of methods that have been developed to compensate in both disease-free and endemic situations. Using FMD as a case example, this review first discusses the different activities that competent authorities use farm-level animal population data for to support (1) preparedness activities in disease-free countries, (2) response activities during an acute outbreak in a disease-free country, and (3) eradication and control activities in an endemic country. We then discuss (4) data requirements needed to support epidemiological investigations, surveillance, and disease spread modelling both in disease-free and endemic countries.

FMD-VS: A virtual sensor to index FMD virus scattering

Foot-and-mouth disease (FMD) models—analytical models for tracking and analyzing FMD outbreaks—are known as dominant tools for examining the spread of the disease under various conditions and assessing the effectiveness of countermeasures. There has been some remarkable progress in modeling research since the UK epidemic in 2001. Several modeling methods have been introduced, developed, and are still growing. However, in 2010 when a FMD outbreak occurred in the Miyazaki prefecture, a crucial problem reported: Once a regional FMD outbreak occurs, municipal officials in the region must make various day-to-day decisions throughout this period of vulnerability. The deliverables of FMD modeling research in its current state appear insufficient to support the daily judgments required in such cases. FMD model can be an efficient support tool for prevention decisions. It requires being conversant with modeling and its preconditions. Therefore, most municipal officials with no knowledge or experience found full use of the model difficult. Given this limitation, the authors consider methods and systems to support users of FMD models who must make real-time epidemic-related judgments in the infected areas. We propose a virtual sensor, designated “FMD-VS,” to index FMD virus scattering in conditions where there is once a notion of FMD; and (2) shows how we apply the developed FMD-VS technique during an outbreak. In (1), we show our approach to constructing FMD-VS based on the existing FMD model and offer an analysis and evaluation method to assess its performance. We again present the results produced when the technique applied to 2010 infection data from the Miyazaki Prefecture. For (2), we outline the concept of a method that supports the prevention judgment of municipal officials and show how to use FMD-VS.

Realistic assumptions about spatial locations and clustering of premises matter for models of foot-and-mouth disease spread in the United States

Spatially explicit livestock disease models require demographic data for individual farms or premises. In the U.S., demographic data are only available aggregated at county or coarser scales, so disease models must rely on assumptions about how individual premises are distributed within counties. Here, we addressed the importance of realistic assumptions for this purpose. We compared modeling of foot and mouth disease (FMD) outbreaks using simple randomization of locations to premises configurations predicted by the Farm Location and Agricultural Production Simulator (FLAPS), which infers location based on features such as topography, land-cover, climate, and roads. We focused on three premises-level Susceptible-Exposed-Infectious-Removed models available from the literature, all using the same kernel approach but with different parameterizations and functional forms. By computing the basic reproductive number of the infection (R₀) for both FLAPS and randomized configurations, we investigated how spatial locations and clustering of premises affects outbreak predictions. Further, we performed stochastic simulations to evaluate if identified differences were consistent for later stages of an outbreak. Using Ripley’s K to quantify clustering, we found that FLAPS configurations were substantially more clustered at the scales relevant for the implemented models, leading to a higher frequency of nearby premises compared to randomized configurations. As a result, R₀ was typically higher in FLAPS configurations, and the simulation study corroborated the pattern for later stages of outbreaks. Further, both R₀ and simulations exhibited substantial spatial heterogeneity in terms of differences between configurations. Thus, using realistic assumptions when de-aggregating locations based on available data can have a pronounced effect on epidemiological predictions, affecting if, where, and to what extent FMD may invade the population. We conclude that methods such as FLAPS should be preferred over randomization approaches.

When modeling the spread of infectious livestock diseases such as foot-and-mouth disease (FMD), the distance between premises is an important aspect. In the U.S., locations of premises are not available, forcing modelers to make assumptions about their coordinates. Such assumptions can be more or less crude and will impact the conclusions drawn from the model. To investigate the impact of such assumptions, we modeled outbreaks of FMD within the cattle population of the U.S. under two assumptions about premises locations. Their position was either randomly distributed within counties or informed by a state-of-the-art method developed specifically to simulate realistic locations of agricultural operations. We found that the higher degree of spatial clustering of premises associated with more realistic assumptions about locations leads to a substantially higher risk of outbreaks. Our results also show that the amount with which the risk is under-estimated by randomizing locations is unevenly distributed across the landscape. Together, these findings show a clear support for using informed methods to determine the spatial locations of premises and highlight the importance of spatial clustering when modeling FMD-like diseases.

Environmental persistence of foot-and-mouth disease virus applied to endemic regions

The consequences of foot-and-mouth disease impact regional economies and food security through animal mortality and morbidity, trade restrictions and burdens to veterinary infrastructure. Despite efforts to control the disease, some regions, mostly in warmer climates, persistently report disease outbreaks. Consequently, it is necessary to understand how environmental factors influence transmission, of this economically devastating disease. Extensive research covers basic aetiology and transmission potential of livestock and livestock products for foot-and-mouth disease virus (FMDV), with a subset evaluating environmental survival. However, this subset, completed in the early to mid-20th century in Northern Europe and the United States, is not easily generalized to today's endemic locations. This review uncovered 20 studies, to assess current knowledge and analyse the effects of environmental variables on FMDV survival, using a Cox proportional hazards (Coxph) model. However, the dataset is limited, for example pH was included in three studies and only five studies reported both relative humidity (RH) and temperature. After dropping pH from the analysis, our results suggest that temperature alone does not describe FMDV survival; instead, interactions between RH and temperature have broader impacts across various conditions. For instance, FMDV is expected to survive longer during the wet season (survival at day 50 is ~90% at 16°C and 86% RH) versus the dry season (survival at day 50 approaches 0% at 16°C and 37.5% RH) or comparatively in the UK versus the Southwestern United States. Additionally, survival on vegetation topped 70% on day 75 when conditions exceeded 20°C with high RH (86%), drastically higher than the survival on inanimate surfaces at the same temperature and RH (~0%). This is important in tropical regions, where high temperatures can persist throughout the year, but RH varies. Therefore, parameter estimates, for disease modelling and control in endemic areas, require environmental survival data from a wider range of conditions.

Challenges and Opportunities Developing Mathematical Models of Shared Pathogens of Domestic and Wild Animals

Diseases that affect both wild and domestic animals can be particularly difficult to prevent, predict, mitigate, and control. Such multi-host diseases can have devastating economic impacts on domestic animal producers and can present significant challenges to wildlife populations, particularly for populations of conservation concern. Few mathematical models exist that capture the complexities of these multi-host pathogens, yet the development of such models would allow us to estimate and compare the potential effectiveness of management actions for mitigating or suppressing disease in wildlife and/or livestock host populations. We conducted a workshop in March 2014 to identify the challenges associated with developing models of pathogen transmission across the wildlife-livestock interface. The development of mathematical models of pathogen transmission at this interface is hampered by the difficulties associated with describing the host-pathogen systems, including: (1) the identity of wildlife hosts, their distributions, and movement patterns; (2) the pathogen transmission pathways between wildlife and domestic animals; (3) the effects of the disease and concomitant mitigation efforts on wild and domestic animal populations; and (4) barriers to communication between sectors. To promote the development of mathematical models of transmission at this interface, we recommend further integration of modern quantitative techniques and improvement of communication among wildlife biologists, mathematical modelers, veterinary medicine professionals, producers, and other stakeholders concerned with the consequences of pathogen transmission at this important, yet poorly understood, interface.

Does Size Matter to Models? Exploring the Effect of Herd Size on Outputs of a Herd-Level Disease Spread Simulator

Disease spread modeling is widely used by veterinary authorities to predict the impact of emergency animal disease outbreaks in livestock and to evaluate the cost-effectiveness of different management interventions. Such models require knowledge of basic disease epidemiology as well as information about the population of animals at risk. Essential demographic information includes the production system, animal numbers, and their spatial locations yet many countries with significant livestock industries do not have publically available and accurate animal population information at the farm level that can be used in these models. The impact of inaccuracies in data on model outputs and the decisions based on these outputs is seldom discussed. In this analysis, we used the Australian Animal Disease model to simulate the spread of foot-and-mouth disease seeded into high-risk herds in six different farming regions in New Zealand. We used three different susceptible animal population datasets: (1) a gold standard dataset comprising known herd sizes, (2) a dataset where herd size was simulated from a beta-pert distribution for each herd production type, and (3) a dataset where herd size was simplified to the median herd size for each herd production type. We analyzed the model outputs to compare (i) the extent of disease spread, (ii) the length of the outbreaks, and (iii) the possible impacts on decisions made for simulated outbreaks in different regions. Model outputs using the different datasets showed statistically significant differences, which could have serious implications for decision making by a competent authority. Outbreak duration, number of infected properties, and vaccine doses used during the outbreak were all significantly smaller for the gold standard dataset when compared with the median herd size dataset. Initial outbreak location and disease control strategy also significantly influenced the duration of the outbreak and number of infected premises. The study findings demonstrate the importance of having accurate national-level population datasets to ensure effective decisions are made before and during disease outbreaks, reducing the damage and cost.

Preserving privacy whilst maintaining robust epidemiological predictions

Mathematical models are invaluable tools for quantifying potential epidemics and devising optimal control strategies in case of an outbreak. State-of-the-art models increasingly require detailed individual farm-based and sensitive data, which may not be available due to either lack of capacity for data collection or privacy concerns. However, in many situations, aggregated data are available for use. In this study, we systematically investigate the accuracy of predictions made by mathematical models initialised with varying data aggregations, using the UK 2001 Foot-and-Mouth Disease Epidemic as a case study. We consider the scenario when the only data available are aggregated into spatial grid cells, and develop a metapopulation model where individual farms in a single subpopulation are assumed to behave uniformly and transmit randomly. We also adapt this standard metapopulation model to capture heterogeneity in farm size and composition, using farm census data. Our results show that homogeneous models based on aggregated data overestimate final epidemic size but can perform well for predicting spatial spread. Recognising heterogeneity in farm sizes improves predictions of the final epidemic size, identifying risk areas, determining the likelihood of epidemic take-off and identifying the optimal control strategy. In conclusion, in cases where individual farm-based data are not available, models can still generate meaningful predictions, although care must be taken in their interpretation and use.

Simulating the Distribution of Individual Livestock Farms and Their Populations in the United States: An Example Using Domestic Swine (Sus scrofa domesticus) Farms

Livestock distribution in the United States (U.S.) can only be mapped at a county-level or worse resolution. We developed a spatial microsimulation model called the Farm Location and Agricultural Production Simulator (FLAPS) that simulated the distribution and populations of individual livestock farms throughout the conterminous U.S. Using domestic pigs (Sus scrofa domesticus) as an example species, we customized iterative proportional-fitting algorithms for the hierarchical structure of the U.S. Census of Agriculture and imputed unpublished state- or county-level livestock population totals that were redacted to ensure confidentiality. We used a weighted sampling design to collect data on the presence and absence of farms and used them to develop a national-scale distribution model that predicted the distribution of individual farms at a 100 m resolution. We implemented microsimulation algorithms that simulated the populations and locations of individual farms using output from our imputed Census of Agriculture dataset and distribution model. Approximately 19% of county-level pig population totals were unpublished in the 2012 Census of Agriculture and needed to be imputed. Using aerial photography, we confirmed the presence or absence of livestock farms at 10,238 locations and found livestock farms were correlated with open areas, cropland, and roads, and also areas with cooler temperatures and gentler topography. The distribution of swine farms was highly variable, but cross-validation of our distribution model produced an area under the receiver-operating characteristics curve value of 0.78, which indicated good predictive performance. Verification analyses showed FLAPS accurately imputed and simulated Census of Agriculture data based on absolute percent difference values of < 0.01% at the state-to-national scale, 3.26% for the county-to-state scale, and 0.03% for the individual farm-to-county scale. Our output data have many applications for risk management of agricultural systems including epidemiological studies, food safety, biosecurity issues, emergency-response planning, and conflicts between livestock and other natural resources.

An approach for de-identification of point locations of livestock premises for further use in disease spread modeling

We describe a method for de-identifying point location data used for disease spread modeling to allow data custodians to share data with modeling experts without disclosing individual farm identities. The approach is implemented in an open-source software program that is described and evaluated here. The program allows a data custodian to select a level of de-identification based on the K-anonymity statistic. The program converts a file of true farm locations and attributes into a file appropriate for use in disease spread modeling with the locations randomly modified to prevent re-identification based on location. Important epidemiological relationships such as clustering are preserved to as much as possible to allow modeling similar to those using true identifiable data. The software implementation was verified by visual inspection and basic descriptive spatial analysis of the output. Performance is sufficient to allow de-identification of even large data sets on desktop computers available to any data custodian.

Eight challenges in modelling infectious livestock diseases

The transmission of infectious diseases of livestock does not differ in principle from disease transmission in any other animals, apart from that the aim of control is ultimately economic, with the influence of social, political and welfare constraints often poorly defined. Modelling of livestock diseases suffers simultaneously from a wealth and a lack of data. On the one hand, the ability to conduct transmission experiments, detailed within-host studies and track individual animals between geocoded locations make livestock diseases a particularly rich potential source of realistic data for illuminating biological mechanisms of transmission and conducting explicit analyses of contact networks. On the other hand, scarcity of funding, as compared to human diseases, often results in incomplete and partial data for many livestock diseases and regions of the world. In this overview of challenges in livestock disease modelling, we highlight eight areas unique to livestock that, if addressed, would mark major progress in the area.