Simulating the transmission and control of highly pathogenic avian influenza in the Australian commercial poultry industries

Abstract

Although Australia is free from highly pathogenic avian influenza (HPAI), incursions of this disease have the potential to have severe impacts on animal welfare and cause substantial economic losses for Australia’s commercial poultry industries. Australia has experienced five epidemics of HPAI between 1976 and 1997, and has detailed contingency plans for this disease, but the potential consequences and effectiveness of different control and eradication strategies under Australian conditions are not well understood. Epidemiological modelling is a tool that can be used to combine available information on infection dynamics from field and experimental studies with expert opinion to gain insights into disease transmission and control. This thesis describes the application of modelling tools to study the spread of HPAI within and between farms. In particular, it describes the development, verification, validation, sensitivity analysis and evaluation of a spatial, stochastic simulation model, called AISPREAD, to investigate the potential transmission and control of HPAI in the Australian chicken meat, chicken egg layer, duck and turkey industries. AISPREAD is divided into three modules: a production module to simulate the timing of farm husbandry practices such as routine destocking and visits by service providers; an infection module to simulate the spread of infection between farms by local spread, the movement of live birds and contact networks of service providers; and a mitigations module which implements combinations of realistic outbreak response strategies. AISPREAD is a flexible simulation model that can incorporate different combinations of mitigation strategies including: active and passive surveillance; diagnosis; tracing; zoning; quarantine and movement restrictions; culling, disinfection and decontamination of infected farms; pre-emptive culling; process slaughter of chicken meat broiler and duck and turkey grower farms; and emergency vaccination. Data on the location, ownership, type, capacity, biosecurity and production characteristics and identity of service providers were obtained from a series of cross-sectional surveys of the commercial chicken meat, chicken egg layer, duck and turkey industries conducted in 2005 and 2007. These studies revealed that these industries are functionally distinct and few of the 1594 farms produced multiple species of poultry. However, five Australian poultry production regions had poultry farm densities that were comparable to areas in Canada and Italy that have experienced extensive epidemics of HPAI, indicating that they may be more vulnerable to larger outbreaks. Results also revealed that movements of birds and service providers between farms could lead to the transmission of HPAI between companies, regions or industries. Limited data were available to estimate several important model parameters, including the likelihood of HPAI spreading between farms by different types of indirect contacts; the relative effectiveness of biosecurity measures on different types of farms to prevent indirect transmission of HPAI on fomites; the number of trucks operated by feed mills, hatcheries and waste disposal companies; and the risk of HPAI spread posed by the movement of used cardboard egg trays in the chicken egg layer industry. A modified Delphi conference approach was used in this study to gather the opinions of Australian poultry industry professionals to estimate these parameters. This study confirmed that poultry industry experts considered that indirect contacts by service providers between poultry farms and differences in biosecurity practices between poultry farms would be important in the spread of HPAI amongst poultry farms. In order to investigate within-farm spread of HPAI, a stochastic discrete entity simulation model was developed to estimate parameters regarding the dynamics of HPAI in unvaccinated and vaccinated chicken meat, chicken egg layer, duck and turkey farms. These studies revealed that a 1% mortality threshold for reporting suspected HPAI in unvaccinated farms would allow for the reporting of infection within one to two weeks after introduction. Vaccination was also shown to reduce the susceptibility of chicken, turkey and duck farms to incursions of HPAI. Two-dose vaccination regimes prevented between 25 and 38% of incursions becoming established in chicken and turkey farms, and hence could reduce the overall susceptibility of chicken and turkey farms to infection. Sentinel surveillance programs, where unvaccinated chickens are placed in flocks and monitored for clinical signs of HPAI, were shown to be beneficial in the detection of HPAI epidemics in vaccinated chicken meat, chicken egg layer, turkey and duck farms. AISPREAD was used to study the transmission of HPAI between farms on a national scale. It performed well in operational validation studies: model outputs were comparable to the 1997 epidemic of HPAI in Tamworth, NSW (New South Wales) and the simulated reproduction ratio (R0) of farms in the Sydney region was similar to those experienced in overseas epidemics in regions in Italy in 1999/2000, the Netherlands in 2003 and Canada in 2004 with similar poultry farm densities. Sensitivity analysis studies of AISPREAD indicated that the model was relatively robust to many uncertain parameters. Further, results were influenced by assumptions regarding the local spread of infection, the sensitivity of surveillance visits and tracing, and the likelihood of transmission by certain indirect contacts (the movement of used egg trays, feed deliveries, litter and manure removal contractors and slaughter crews). Although these parameters were estimated using the best available data, better estimates would improve the validity of AISPREAD. Studies were conducted using AISPREAD to assess the potential consequences of HPAI incursions (measured by the number of infected farms, epidemic duration, size of Restricted Areas and the number of dead and destroyed birds) into different types of poultry farms across different poultry production regions in Australia. In these studies it was assumed that a basic stamping-out policy would be implemented (involving the quarantine and the culling of infected farms, the imposition of zoning and movement restrictions and active surveillance on at risk farms). Incursions into chicken meat broiler farms in areas where there are extensive indirect contacts, such as the Central Coast of NSW and the Mornington Peninsula of Victoria, tended to result in larger epidemics compared with epidemics seeded into turkey grower, duck grower, chicken meat parent breeder and combined chicken egg layer and pullet farms. A basic stamping-out strategy was successful in preventing the secondary transmission of HPAI in most incursion scenarios studied. In most cases, AISPREAD forecast that HPAI would be eradicated in less than three weeks, resulting in less than 200,000 birds dying from infection or being destroyed. The effectiveness of several alternative control strategies for HPAI (including stamping-out within larger disease control zones, pre-emptive ring culling, process slaughter, and emergency vaccination) were also assessed for epidemics where HPAI was introduced into a chicken meat broiler farm on the Mornington Peninsula. This study showed that mitigation strategies involving larger disease control zones may be more effective in controlling HPAI epidemics in regions considered most vulnerable to the secondary spread of infection between farms. However, unless strategies were put in place to limit the impact of movement restrictions on chicken meat broiler and duck and turkey grower farms, larger disease control zones would lead to a substantial number of farms (median 23, 95% probability interval, 2 to 122 farms) being overdue to move birds to slaughter, which could result in animal welfare problems. A process slaughter strategy, where chicken meat broiler and duck and turkey grower farms in the vicinity of infected farms were processed after testing negative for HPAI, reduced the number of farms disrupted by movement restrictions and did not result in larger epidemics. Results of this study indicate that emergency vaccination may offer only limited benefits compared with other strategies, at least for moderately sized epidemics. However, should government and industry decide to retain emergency vaccination as a contingency, 4.6 million doses of AI vaccine would be required within two months to manage an incursion in a high poultry density region. The approach used in AISPREAD to capture production events and periodic contacts by service providers between commercial intensive poultry farms has allowed the protective effects of the periodic depopulation and restocking of single-aged poultry farms to be captured and a comprehensive list of transmission pathways to be included in the model. Similar approaches may be useful to represent the transmission of other important poultry diseases or diseases in other intensive industries, such as the pig industry, where farms are managed in an all-in all-out manner and indirect contacts by service providers can lead to the transmission of infection. It is recommended that AISPREAD be updated in the future as more becomes understood about the epidemiology of HPAI and as the structure and dynamics of the poultry industries in Australia change.