Risk-based surveillance of avian influenza in switzerland : insights into poultry farm determinants and contact networks

This dissertation explores new potentials for a risk-based surveillance of highly pathogenic avian influenza (HPAI) in domestic poultry in an outbreak-free situation in Switzerland, 2007 to 2009. Its focus is on risk factors for HPAI spread related to the host population, namely poultry farms. In Part 1, poultry keepers’ disease awareness, potential contagious contacts amongst poultry farms, and the issue of organizing basic demographic data are addressed. In Part 2, emphasis is placed on the interplay between population data, namely contact structures, and mathematical models of epidemics, in general, and on the specific case of the Swiss poultry population.
Part1
Using a mixed methods research design, consisting of a countrywide cross-sectional survey among 3,978 poultry keepers and interviews with experts from the poultry industry, revealed that: (i) risks perceived by the poultry keepers well reflected the officially communicated ones; (ii) having a non-commercial poultry farm was associated with low knowledge scores; (iii) both commercial and non-commercial farms were involved with remote between-farm contacts, which was in contrast to commonly presumed small poultry movement ranges in the non-commercial sector. Organizing demographic and epidemiological information in a relational poultry farm database would allow for considering more than one farm site per poultry keeper and facilitate multi-criteria queries.
Part 2
Computer simulations comparing two types of individual-based models for the total outbreak size showed that: (i) models assuming a randomly mixed population provided acceptable estimates of the total outbreak size if the number of contacts per day was high or if the per-contact transmission probability was high; as well as for diseases with very short infectious periods; yet (ii) models considering the structure and stability of potentially contagious contacts were superior if the number of daily contacts or the transmission probability were low.
Extrapolating the measured contact profiles on all poultry farms in Switzerland resulted in a synthetic poultry farm population characterized by heterogeneously distributed contacts with most farms having no or one contact and 4% of the farms, including non-commercial ones, having 4 or more different contacts.
Overall, a particular need was identified to integrate non-commercial farms in HPAI surveillance. Surveillance intensity would ideally be based on multiple criteria risk weighting and rating, and always be high in poultry and farm dense areas. Epidemic models at the between-farm level for HPAI should take the realistic arrangement of contacts into account. A contact network model for the Swiss poultry farm population is feasible although computationally and labor intensive. A complete and flexible poultry farm database is required for both, targeting surveillance and epidemic modeling. While this dissertation was written the legal basis for a complete poultry register on a federal level has been created in Switzerland.