Within the European Union, epidemics of contagious animal diseases such as Classical Swine Fever (CSF) and Foot-and-Mouth Disease (FMD) are to be eradicated according to strict EU- prescriptions including stamping-out of infected herds, establishment of control and surveillance zones with complete standstill of animals and possible export bans on live animals. Epidemics clearly have a serious impact, in particular on countries with a high farm density and an export- oriented production, such as the Netherlands. Therefore, an effective animal disease prevention policy is of major importance for these countries.
This thesis is a result of a joint action by the government and private industry in the Netherlands that have funded a research project aimed at gaining more insight into the risk and economic consequences of virus introduction into the country.
Real-life experiments on strategies to reduce animal disease introduction and spread is not an option because such experiments would be too risky (and hence too costly). In principle, simulation modelling is an attractive alternative. It calculates the effects of predefined sets of input variables and therefore also offers the possibility of exploring strategies that have not been applied yet. Literature search showed that simulation models describing spread and economic consequences of epidemics within a country were already available. However, an integrated approach which combines the various aspects of outbreaks and spread with economic consequences of outbreaks was still lacking. Therefore, this research project emphasized the development of a model describing introduction of virus into the Netherlands and on the integration and combination of the models.
As with every model, the quality of the outcome of a simulation model is strongly influenced by the quality of the input ('garbage in = garbage out'). Therefore, a considerable part of the research was devoted to the gathering of information on aspects influencing virus introduction. It would be ideal to base the simulation model on relevant historical and experimental data. However, such information on virus introduction is limited, if available at all. Furthermore, outbreaks of contagious animal diseases occur irregularly over time and differ in magnitude; moreover circumstances change. Therefore it is questionable if historical data are relevant in simulating current and future events. Experimental data are also sparsely available. Literature search has shown that many researchers have done work in the area of contagious animal diseases, but most of their findings are of a qualitative nature. Despite this lack of 'objective' information, decisions on eradication and prevention of outbreaks must be made. Currently, such decisions rely on the expertise (a combination of experience and understanding of current/future circumstances) of those working in this area. Such expertise is a useful and necessary addition to the data available from research and databases. The elicitation of this 'expert knowledge' in an objective way, resulting in quantitative information useful for modelling purposes, was one of the major topics in the thesis.
For the elicitation experiment a format was sought which would guarantee a high response from experts and would provide the ability to elicit individual opinions in an objective way. Literature search showed that many elicitation methods were available, all with their own pros and cons. One of the methods, Conjoint Analysis, was considered an interesting technique for the elicitation of the relative importance of the risk factors. Conjoint Analysis is well known in consumer and marketing studies, but thus far has not or only scarcely been used in the field of animal health economics. A pilot experiment was conducted in which the potential of the method for elicitation of the relative importance of risk factors was explored. In this experiment, the Conjoint Analysis technique was used to draw up a paper questionnaire which was handed out during the 7th International Symposium on Veterinary Epidemiology and Economics (ISVEE) held at Nairobi, Kenya, August 1994. Relevant ISVEE-participants asked to assign scores to profiles of six contagious animal diseases (African and Classical Swine Fever, Foot-and-Mouth Disease, Swine Vesicular Disease, Newcastle Disease, and Avian Influenza). The Kenya-experiment showed promising results. However, also some lessons for future use were learned. The 'paperapproach' resulted in a low response (30%) and the participants, who were not selected on the basis of their expertise, indicated that is was very difficult and time-consuming to evaluate six diseases at one time.
Basic experiment: elicitation of opinion of Dutch experts
The results and experiences of the Kenya experiment have led to and framed the use 0' Conjoint Analysis in a second and much bigger experiment during which the subjective knowledge of Dutch experts was elicited. Experts were defined as people that were involved in either research or policy-making on animal disease prevention and people that would be consulted in case of an outbreak. As the number of experts was limited, an approach was needed which would guarantee a high response rate. Therefore, the experiment was framed as a full evening's workshop. Such a workshop is a one-off group meeting, which means that the participants have to show up only once and have the possibility of discussing issues with other experts: both aspects may be attractive and incentives to join. This approach worked out well: the experiment was attended by a total of 43 out of 50 invited persons, a response rate of 86%.
Although it was acknowledged that group discussions may have the advantage of resulting in new and better approaches because people are able to share, evaluate and stimulate each other's opinions, the risk of possible negative effects of such discussions, such as individual dominance, was too high. Therefore, the workshop participants were asked to individually complete a computerized questionnaire. The program was developed to be self-explanatory, which minimized the interaction among participants and between participants and facilitators. The participants were given the opportunity to indicate on which disease they felt themselves most knowledgeable and were only asked questions about that disease. Furthermore, in order to keep the whole exercise at a manageable size, the questions were confined to Europe, the countries being grouped into five clusters.
Relative importance of risk factors
The relative importance of risk factors responsible for the introduction of virus into the Netherlands was elicited by using the above-described Conjoint Analysis method. Questions were asked per country cluster. The results showed that, for both FMD and CSF major risk factors were import of livestock and returning trucks. Differences between country clusters were small. For NCD, major risk factors were import of live animals, transport materials (crates) and import of exotic birds.
As a follow-up on the Kenya experiment, the Conjoint Analysis element of the workshop experiment was evaluated in detail on its usefulness as a tool to elicit expert knowledge in the field of animal health economics. Criteria were validity, consistency and respondent evaluation. The results obtained were comparable to or better than the results obtained in consumer and marketing studies. It was concluded that Conjoint Analysis provided a useful addition to the toolkit of the animal health economist.
Frequency of outbreaks in Europe
Each time an outbreak occurs in a European country, there is always a chance that the virus will be transferred to the Netherlands. A higher frequency of outbreaks in Europe means a higher risk of virus introduction to the Netherlands (and to all other countries). It was expected that estimates on the frequency of outbreaks would be difficult to make. Therefore, a method was chosen which enabled expression of uncertainty. This method, called ELI (elicitation), is a graphically-oriented computer program which facilitates the quantification of subjective knowledge about uncertain quantities. The program helps respondents produce subjective probability density functions (PDFs) and is based on socalled proper scoring rules. The top of a PDF indicates the best guess or most likely value, according to the respondent. The dispersion corresponds with the uncertainty as to this estimate.
The ELI-element of the workshop resulted in the parameters of a normal distribution from all participants individually concerning the expected number of outbreaks within the next five years. The workshop participants expected numerous outbreaks of NCD and CSF but not so many of FMD. For all three diseases, most outbreaks were expected to occur in Eastern Europe (midpoints were 21, 20, and 21, for CSF FMD and NCD respectively). The smallest numbers were expected in the country cluster containing the UK, Ireland and Scandinavia (midpoints were 1, 0.5, and 2.5, for CSF, FMD and NCD respectively).
High Risk Period
The High Risk Period (HRP) defines the period in which the virus is already present in a country but not yet detected or under control. During this period, the virus may spread freely within the country and/or transferred to other countries. The HRP can be divided into two periods, the first one starting with infection of the first animal and ending when an infected animal is detected (HRP1), the second period starting with detection and ending when all measures are considered effective (i.e., no spread to other countries) (HRP2).
The participating experts were asked to give a three-point estimate (minimum, most likely and maximum expected length) for the duration of both periods, for all country clusters and for the Netherlands. The HRP Is with the highest duration were expected for Eastern Europe (midpoints of 42, 19, and 21 days for CSF, FMD and NCD respectively). Shortest duration was expected for the cluster including the UK, Ireland and Scandinavia (midpoints were 21, 7, and 10 days for CSF, FMD and NCD respectively). Short durations were estimated for the Netherlands as well.
The HRP2 estimates showed the same distribution over countries, but were longer for FMD and CSF and shorter for NCD.Edit abstract