COMBINING QUANTITATIVE RISK ASSESSMENT AND SYNDROMIC SURVEILLANCE

Get Complete Project Material File(s) Now! »

Equine industry, risk and specificities

Population at risk

Horses, unlike livestock, typically travel frequently over short and long distances around the world for competition, training, and/or reproduction. These movements increase the risk of the dissemination of infectious diseases (Robin et al. 2011). This is a concern not just for the equine industry but also for public health. Indeed, numerous equine viruses are zoonotic (e.g., rabies, brucellosis, anthraw, glanders, leptospirosis, Hendra virus). However, controlling diseases spread by equines is not only important from a sanitary point of view but also with regard to the important economic weight of the equine industry, particularly in Europe (Liljenstolpe 2009). As an example, in 2010, the European equine industry encompassed 3.7 million horses, generated 100 billion euros a year, and provided the equivalent of 400,000 full-time jobs (Leadon and Herholzt 2009). Furthermore, the sector is growing, with an increase in the number of horse riders of 5% per annum. The introduction of exotic infectious disease may thus have huge economic consequences, as was seen with the 13-week outbreak of African horse sickness in Portugal in 1990, whose total cost was estimated around US $2 million (Portas et al. 1999).

Population not well tracked

Despite the sanitary and economic impacts of equine diseases, effective health regulations and biosecurity systems to ensure safe equine movements are not always in place at the national and international level (Leadon and Herholzt 2009; Murray et al. 2013). This was illustrated by the outbreak of equine influenza in Australia in 2007. Here, the authorities failed to contain the infection in quarantine following the importation of one or more infected horses (Webster 2011). The horse population is also not well-tracked, which complicates the control and surveillance of diseases. In the EU, the implementation of mandatory passports for horses in 2008 has improved the tracking of horses. However, the database that contains the information on animal movements and deaths is not regularly updated. This has two consequences. Firstly, the exact number of horses and their geographical location is unknown, which is an obstacle for disease surveillance and control. Secondly, the exact number of horses transported between EU member states or within a country is still not available. Indeed, although the EU’s Trade Control and Expert System (TRACES) (Commission Decision 2003) provides information on the number of horses imported to and within the EU, several movements are not recorded in the database due to the absence of mandatory transport notification.
The difficulties of implementing proper health regulations and tracking systems can be explained by the complex reality of the world of horses. The equine industry includes a myriad of activities (e.g., tourism, equestrian sports, breeding and slaughtering of horses) and the various stakeholders engaged in each activity do so with different expectations, ranging from professional to leisure (Castejón-Montijano and Rodríguez-Fernández 2011).
Considering the potential health and financial risks posed by horses, it is especially important to develop novel approaches for the surveillance of exotic infectious diseases, such as VBDs. However, this also constitutes an additional challenge given the structure of the equine industry.

Early warning

A key point in controlling emerging or reemerging VBDs is early warning. Indeed, dealing with a disease outbreak in its early stages is easier and more economical than once it has become widespread (FAO-OIE-WHO Collaboration 2013; FAO-OIE-WHO Collaboration 2006). Early warning systems are timely surveillance systems aimed at predicting the probability that an outbreak is spreading to new areas in order to trigger prompt public health interventions (FAO-OIE-WHO Collaboration 2006). Different strategies such as active and/or passive surveillance are used to ensure the timeliness of detection.

Active surveillance

Active surveillance refers to the active role of health authorities in data collection. The advantage is that active sampling may detect a disease without the observation of clinical signs. However, one of the major drawbacks is that, to detect rare diseases like a newly introduced exotic disease, active sampling has to be very large and redundant, which can be very costly (Doherr and Audigé 2001). To mitigate costs, it is possible to implement a specific type of active surveillance, known as risk-based surveillance. Risk-based surveillance is defined by Stärk and colleagues (Stärk et al. 2006) as the allocation of surveillance activities based on the probability of events with or without consideration of the consequences of the event, the management of the event, or the process of communication of the event. The term “targeted surveillance” is also used (Doherr and Audigé 2001; de Koeijer et al. 2002; Salman 2003). The simple idea behind the concept is to collect samples from the segments of the population that have the highest probabilities of being infected, thus increasing the probability of disease detection (Salman 2003). In this way, this process decreases the required sample size without reducing the probability of detecting the disease. Risk-based surveillance is based on the concept of looking for something where it is most likely to be found; this approach thus requires reliable and thorough prior information on at-risk populations in order to ensure the appropriate representativeness of the sampling (FAO 2014; Oidtmann et al. 2013; Stärk et al. 2006).

Passive surveillance

In many countries, passive surveillance is one of the most common forms of surveillance for rare and exotic diseases. The term refers to any passive disease reporting systems in which veterinarians, farmers, or any other stakeholders notify authorities when they have sick animals. These surveillance systems are used to identify numerous diseases since they have several significant advantages: they cover a large part of the animal population and the costs associated with data collection and analysis are relatively low (Doherr and Audigé 2001; FAO 2014; Salman 2003). However, the performance of passive surveillance systems suffers from frequent under-reporting due to the lack of stakeholder awareness regarding a disease of interest. This may result in a failure to identify the disease (Hadorn et al. 2008) especially when it manifests in few or unspecific clinical signs (Doherr and Audigé 2001). In addition, potential fears of the disease’s consequences may also incite stakeholders to not report suspected cases (FAO 2014; Salman 2003). Under-reporting is especially problematic regarding the surveillance of exotic diseases, as exotic diseases have a low probability of occurrence, their symptoms usually not well known by practitioners, and the consequences of reporting an exotic disease in a new area may be dramatic.

Early warning in horse population

For the early detection of exotic VBDs in horses, both active and passive approaches could theoretically be implemented. However, as previously highlighted, one of the major drawbacks of the active approach is that it can be very costly, especially when a disease is rare like exotic VBDs (Doherr and Audigé 2001). Risk-based active sampling could be implemented, but the equine industry suffers from a lack of accurate data regarding the populations at risk (e.g., details of animal movements, population size and location) which might complicate the planning and the implementation of such an active surveillance system. The efficiency of the classical passive surveillance approach in the early detection of an outbreak may also be limited due to the high probability of under-reporting, especially for exotic diseases.
Instead of relying on classical active or passive surveillance to detect new outbreaks, then, new approaches for estimating the probability of outbreak occurrence may constitute a promising way to improve the early detection of VBDs in horses. Different approaches can be considered in estimating this probability: classical risk assessment and syndromic surveillance.

READ  PRICE MOMENTUM ANOMALY

Risk assessments

Risk assessments are the component of risk analysis that estimates the risks associated with a hazard, probability of hazard occurrence and its consequences (OIE 2010). Applied to exotic diseases, they are well-known tools for describing the probability of pathogen entry and spread within an area. The probability of entry is defined as the probability that a pathogen enters in a given area, considering all potential pathways of introduction and without considering the later steps of transmission (OIE 2014). The probability of “spread” is a vaguer concept and can include different sub-definitions, as presented by de Vos and colleagues (de Vos et al. 2011): (1) the probability of transmission, which is defined as the probability that the pathogen is able to spread to susceptible hosts in the area at risk, (2) the probability of establishment, which is the probability that the pathogen is able to spread to susceptible hosts and to susceptible vectors given the conditions of introduction, and (3) the probability of spread, which is the probability that the pathogen is able to spread in time and space, considering both local and long-distance dispersal. Regarding the specific issue of early detection, the probability of spread is irrelevant, as it is related more to the assessment of a disease’s impact when early surveillance has already failed to detect and control an outbreak. Conversely, the probabilities of transmission and establishment are especially interesting for early detection as they indicate the time period and the most suitable area for early spread of a pathogen. In particular, the probability of establishment, which takes into account the place and time of entry, is an interesting parameter with which to evaluate the likelihood of an infection actually leading to local spread.
The above approach gives a probability of outbreak occurrence based on risk factors such as the suitability of an environment and climate for disease transmission, or the presence of risky practices (e.g., importation of animals from infected area). It can be used by decision makers for risk mitigation and/or to enhance stake-holders’ awareness of rare or emerging diseases through risk maps, as has already been proposed for some endemic VBDs (e.g., surveillance of West Nile virus in California (Brown 2012)

Syndromic surveillance

To enhance traditional passive surveillance systems, methods based on the analysis of pre-diagnostic and unspecific routinely collected data were developed at the beginning of the 21st century. Such approaches, referred to as syndromic surveillance, aim to identify the early, often weak, signal of an outbreak in the absence of an accurate identification of the disease by medical practitioners or laboratories. There is no single and commonly accepted definition for syndromic surveillance but it is commonly accepted that it focuses on data collected prior to clinical diagnosis or laboratory confirmation (Katz et al. 2011; Shmueli and Burkom 2010). First developed in human medicine, it is now also widely used in veterinary medicine (Dórea et al. 2011); indeed, a recent review by Dupuy and colleagues (Dupuy et al. 2013a) identified at least 27 syndromic surveillance systems or initiatives in 12 European countries. However, regarding horses, few syndromic surveillance initiatives are in place and only two have been explicitly identified: one in UK with Equine quarterly surveillance reports (DEFRA/AHS/BEVA 2015) and another another in The Netherlands with the GD monitor system (Rockx et al. 2006).
Syndromic surveillance provides a risk of outbreak occurrence based on the abnormal evolution of a health-related indicator. Such approaches can be used to rapidly detect a well-known disease or new pathogen without a priori consideration and they thus promise to strengthen surveillance of VBDs in horses. However, because they rely on health-related indicators, syndromic surveillance usually has a low specificity (variations in the indicator might be due to disease or to another event) and it is not able to take into account other epidemiological information available for a disease, such as environmental risk factors.

Table of contents :

CHAPTER I: INTRODUCTION
A. CONTEXT AND NEEDS
1. Vector-borne diseases, a major concern
2. Equine industry, risk and specificities
2.1. Population at risk
2.2. Population not well tracked
3. Early warning
3.1. Active surveillance
3.2. Passive surveillance
3.3. Early warning in horse population
4. Risk assessments
5. Syndromic surveillance
B. RESEARCH QUESTION
C. CASES STUDIES
1. French equine industry
1.1. Equine population
1.2. Organization of the equine industry
1.3. Disease surveillance in French horses
2. Diseases of interest
2.1. African horse sickness
2.2. Equine encephalosis
2.3. West Nile virus
D. OUTLINES OF THE WORK
CHAPTER II: QUANTITATIVE RISK ASSESSMENTS
A. OVERVIEW
1. General principle of risk assessment
1.1. Definitions and objectives
1.2. Method
2. Probability of entry
2.1. Definition
2.2. Estimation for VBDs
3. Probabilities of transmission and establishment
3.1. Probability of transmission
3.2. Probability of establishment
4. Conclusion
B. PROBABILITIES OF ENTRY AND ESTABLISHMENT
1. Introduction
2. Paper 1
C. COMPARISON OF DISEASES RISKS
1. Introduction
2. Method
2.1. Model for risk assessment of viruses entry
2.2. Parameters
2.3. Input data
2.4. What-if scenarios
2.5. Calculations
3. Results
3.1. Spatiotemporal probability of entry
3.2. Sensitivity analysis
3.3. What-if scenarios
4. Discussion
D. DISCUSSION AND CONCLUSION
1. Discussion
2. Conclusion
CHAPTER III: SYNDROMIC SURVEILLANCE
A. OVERVIEW
1. Overall principle of syndromic surveillance
1.1. History
1.2. Definition and objectives
1.3. Overall approach
2. Preliminary steps
2.1. Definition of objectives
2.2. Overview of data available
2.3. Definition of syndrome
3. Data description and preprocessing
3.1. Data description
3.2. Data preprocessing
4. Detection algorithms
4.1. Choice of detection algorithm
4.2. Historical limits
4.3. Control charts
4.4. Regression methods
5. Assessment of performance
5.1. Performance metrics
5.2. Test data
6. Conclusion
B. COMPARISON OF PRE-PROCESSING METHODS
1. Introduction
2. Methods
2.1. Data characterization
2.2. Data pre-processing
2.3. Forecasting
2.4. Detection algorithm
2.5. Quantitative assessment
2.6. Implementation
3. Results
3.1. Baseline characterization
3.2. Smoothing and forecasting
3.3. Outbreak detection
4. Discussion
5. Conclusion
C. VALUE OF EVIDENCE
1. Introduction
2. Paper 2
D. DISCUSSION AND CONCLUSION
1. Discussion
2. Conclusion
CHAPTER IV: MULTIPLE INDICATORS OF RISK
A. MULTISTREAM SYNDROMIC SURVEILLANCE
1. Introduction
2. Paper 3
B. COMBINING QUANTITATIVE RISK ASSESSMENT AND SYNDROMIC SURVEILLANCE
1. Introduction
2. Material and Methods
2.1. Bayesian framework
2.2. Simulated EE outbreaks
2.3. Probability of EE introduction
2.4. Syndromic surveillance of EE
2.5. Presentation of the concept and first feedback
3. Results
3.1. Multivariate syndromic surveillance
3.2. Combining Risk assessment
3.3. Workshop output
4. Discussion and Conclusion
C. DISCUSSION AND CONCLUSION
1. Discussion
2. Conclusion
CHAPTER V: DISCUSSION
1. Early warning system of VBDs in horses
1.1. Quantitative risk assessment
1.2. Syndromic surveillance
1.3. Combining risk
2. Support decision-making
3. Demonstrate freedom of disease
4. Practical implementation of integrated surveillance systems
5. Conclusion
APPENDICES
REFERENCES

GET THE COMPLETE PROJECT

Related Posts