Comparison of alternatives to passive surveillance to detect foot and mouth disease incursions in victoria, australia

Preventive Veterinary Medicine 128 (2016) 78–86 Contents lists available at ScienceDirect Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed Comparison of alternatives to passive surveillance to detect foot and mouth disease incursions in Victoria, Australia M.G Garner a , I.J East a,∗ , T Kompas b , P.V Ha b , S.E Roche a , H.T.M Nguyen b a b Animal Health Policy Branch, Commonwealth Government – Department of Agriculture, GPO Box 858, Canberra, ACT 2601, Australia Crawford School of Public Policy, Crawford Building (132), Lennox Crossing, Australian National University, Canberra, ACT 0200, Australia a r t i c l e i n f o Article history: Received May 2015 Received in revised form 28 January 2016 Accepted 19 April 2016 Keywords: Foot and mouth disease Surveillance Early detection Simulation modelling a b s t r a c t This study aimed to evaluate strategies to enhance the early detection of foot and mouth disease incursions in Australia Two strategies were considered First, improving the performance of the current passive surveillance system Second, supplementing the current passive system with active surveillance strategies based on testing animals at saleyards or through bulk milk testing of dairy herds Simulation modelling estimated the impact of producer education and awareness by either increasing the daily probability that a farmer will report the presence of diseased animals or by reducing the proportion of the herd showing clinical signs required to trigger a disease report Both increasing the probability of reporting and reducing the proportion of animals showing clinical signs resulted in incremental decreases in the time to detection, the size and the duration of the outbreak A gold standard system in which all producers reported the presence of disease once 10% of the herd showed clinical signs reduced the median time to detection of the outbreak from 20 to 15 days, the duration of the subsequent outbreak from 53 to 42 days and the number of infected farms from 46 to 32 Bulk milk testing reduced the median time to detection by two days and the number of infected farms by six but had no impact on the duration of the outbreak Screening of animals at saleyards provided no improvement over the current passive surveillance system alone while having significant resource issues It is concluded that the most effective way to achieve early detection of incursions of foot and mouth disease into Victoria, Australia is to invest in improving producer reporting Crown Copyright © 2016 Published by Elsevier B.V All rights reserved Introduction Foot and mouth disease (FMD) is one of the most infectious diseases of domestic livestock (OIE, 2012) FMD is endemic in twothirds of the world (Grubman and Baxt, 2004; Kompas et al., 2015) where it causes annual losses of US$6.5–21 billion (Knight-Jones and Rushton, 2013) Increasing international movements and trade present an on-going threat to FMD-free countries In the past 15 years, there have been a number of outbreaks of FMD in previously free countries despite the application of stringent quarantine measures These outbreaks resulted in estimated financial losses of more than US$1.5 billion (Knight-Jones and Rushton, 2013) and Abbreviations: CVO, Chief Veterinary Officer; FMD, foot and mouth disease; GSAT, general surveillance assessment tool; IP, infected premises; PCR, polymerase chain reaction ∗ Corresponding author E-mail addresses: iain.j.east@agriculture.gov.au, iain.east@gmail.com (I.J East) http://dx.doi.org/10.1016/j.prevetmed.2016.04.009 0167-5877/Crown Copyright © 2016 Published by Elsevier B.V All rights reserved substantial disruption to the international livestock trade (Blayney et al., 2006) In the face of the continuing threat of FMD introduction, early detection of an incursion is of particular importance because the longer the time to detection and the larger the size of the outbreak at detection, the more difficult is the task of disease eradication (Carpenter et al., 2011; Matthews, 2011) A common form of surveillance used to detect disease incursions is passive surveillance; the observation and reporting of clinical signs of disease in animals by animal health professionals, para-professionals, animal owners, producers, processors and others across the livestock industries (Hoinville, 2011) Key observation points for livestock include the farm, the market/saleyard and the abattoir Passive surveillance tends to detect diseases associated with unusual or obvious clinical signs While it has its limitations in terms of providing representative information on populations, in timeliness of detection and in having poor sensitivity, it can be a very effective method of identifying new and emerging diseases (Langstaff, 2008) Previous qualitative studies have also shown that the time between first clinical appearance of disease and the actual reporting of that M.G Garner et al / Preventive Veterinary Medicine 128 (2016) 78–86 79 Fig Map of Victoria (the study area) showing major dairying areas (shaded areas) and the location of smaller pig farms (•) disease by farmers is often too long, resulting in extensive spread of the disease (Elbers et al., 1999; Elbers et al., 2010) Reporting of disease by producers is limited by a number of factors including inability to recognise the disease (Hopp et al., 2007), the potential deleterious impact of reporting disease on the individual farm through quarantine, stamping out, etc (Elbers et al., 2010) and a lack of trust in government (Palmer et al., 2009; Elbers et al., 2010) Any actions that could overcome these barriers to reporting could well be effective in enhancing the efficacy of passive surveillance for detecting disease incursions The performance and reliability of the passive surveillance ‘system’ in Australia has been of concern for a number of years, largely owing to reductions in government expenditure on agriculture and a reduction in the veterinary services in rural areas (Nairn et al., 1996; Frawley, 2003; Matthews, 2011) Similar concerns have been expressed in other countries including the United States (Bates et al., 2003) and The Netherlands (Klinkenberg et al., 2005) A recent review of Australia’s preparedness for FMD (Matthews, 2011) found that there is a strong possibility that an incursion of FMD may not be readily detected due to a range of factors Modelling studies in Australia (Martin et al., 2015; East et al., 2016) have indicated that expected times to detection would be 20–33 days with an upper 90th confidence interval of 22–47 days, depending on region The predicted delay to detection is similar to those observed in recent outbreaks in other countries (Anderson, 2002; Bouma et al., 2003; Yoon et al., 2013) It is therefore of interest to examine whether potential exists to improve time to detection through introducing active surveillance, using new methods of surveillance or enhancing the existing passive surveillance system Saleyards and markets have been recognised as important amplifying points for FMD because of their potential facilitate rapid spread of infection over wide areas (Mansley et al., 2003; Animal Health Australia 2014a) Active surveillance using realtime detection systems to identify FMD at saleyards was proposed by Bates et al (2003) as a way to prevent dissemination of disease through transport of infected animals away from the saleyard Hernández-Jover et al (2011) estimated the sensitivity of the current surveillance system in place at Australian pig saleyards and abattoirs for detecting FMD as no more than 0.35 indicating that potential for improvement to saleyard surveillance exists New methods of surveillance may arise through development of new technologies and one example of this is the development of bulk milk testing for FMD (Reid et al., 2006; Thurmond and Perez, 2006) Foot and mouth disease virus is detectable in the milk of infected cows for 1–3 days before clinical signs of infection appear (Blackwell et al., 1982; Reid et al., 2006) This observation provides potential to detect dairy cows infected with FMD prior to the appearance of clinical signs; an earlier time point than possible through passive surveillance The tests developed allow for the detection of FMD virus in samples diluted up to in 104 (Reid et al., 2006) and this would allow testing of bulk milk samples after arrival at a milk processing facility Given these developments in PCR diagnostic technology for detecting FMD virus in milk samples, the feasibility of testing milk samples for FMD is of particular interest because sampling schemes to collect bulk milk are already in place for a number of quality assurance programs This paper aims to examine the potential for improving early detection of FMD in Victoria, Australia through: Enhancing the performance of the current passive surveillance system through education campaigns that increase producer awareness of disease and the capacity to recognise diseased animals Active surveillance programs at saleyards Active surveillance using bulk milk testing To so, we used a spatially dynamic epidemiological model for FMD in Australia (Garner and Beckett, 2005) to assess the impact of these programs on the time to detection before an FMD outbreak is reported and the size and duration of the outbreak at the time of reporting Issues influencing the effectiveness of these strategies are also discussed Materials and methods 2.1 Study area The study area for this project is the state of Victoria (Fig 1) where the temperate, climate and higher rainfall allow more intensive farming than much of the rest of Australia Victoria is Australia’s largest food and fibre exporting state and is the centre of Australia’s dairy production It has 9.2% of the national beef cattle population, 63.6% of the dairy cattle population, 24.8% of the pig population and 21.3% of the sheep population (ABARES, 2014) The study area contains 42,279 farms with FMD susceptible species categorised into one of eight different types (see below) for the purposes of 80 M.G Garner et al / Preventive Veterinary Medicine 128 (2016) 78–86 Table The number of farms by type in the study area Code Farm type Count Beef Dairy Sheep Pig Beef-sheep Smallholder Feedlot/finisher Total 7537 7590 4017 322 5392 17,233 188 42,279 modelling (Table 1) Dairy production is concentrated in three discrete areas of the state (Fig 1) Because of the mild climate, higher stocking rates, relatively high human population density and proximity to airports and seaports, Victoria is considered to be a higher risk area of Australia for FMD introduction, establishment and spread (East et al., 2013) 2.2 Disease spread modelling The Australian Department of Agriculture’s FMD model AusSpread was used to simulate outbreaks of FMD AusSpread (Garner and Beckett, 2005; East et al., 2014a,b; Garner et al., 2014) is a stochastic spatial simulation model in which FMD transmission is modelled through five discrete pathways: farm to farm animal movements, local spread (infection of farms within close geographical proximity by unspecified means), indirect contact (via contaminated fomites or animal products), animal movements via sale-yards or markets and wind-borne spread Eight different farm types are recognised in the model including specialist beef, dairy, sheep, small pig, large pig, mixed beef and sheep, feedlot producers and small-holders Small-holders are small land owners with livestock, who are often sub commercial and have lifestyle as a key motivator The model incorporates the attributes and spatial locations of individual farms, saleyards, weather stations, local government areas and various other features of the regional environment AusSpread is configured to support the range of control measures described in Australia’s emergency animal disease response arrangements (Animal Health Australia, 2014a) The disease control options available in the model include a national livestock standstill period and subsequent movement restrictions in Restricted Areas (a relatively small legally declared area around infected premises and dangerous contact premises that is subject to disease controls, including intense surveillance and movement controls) and Control Areas a legally declared disease-free buffer area between the Restricted Area and uninfected areas (Animal Health Australia, 2014b) around infected premises (IPs), surveillance and tracing, and stamping out of IPs Pre-emptive culling of properties around IPs and/or traced dangerous contact premises or vaccination may also be considered To support this study, an additional module was added to AusSpread to simulate detection of FMD through the passive surveillance system This module has now been incorporated into AusSpread so the user has a choice of determining when the first detection after an incursion will be made – either fixed time or through passive reporting Passive reporting was implemented by applying probabilities that suspect FMD, based on clinical signs of disease, would be reported by livestock owners, inspectors at saleyards or inspectors at abattoirs The probability of reporting was conditional on a minimum clinical prevalence on an infected farm A series of time steps to allow for investigation, confirmation and reporting to the Chief Veterinary Officer (CVO), were added to determine when an emergency disease response would be triggered The approach is described in more detail below Table The time delay until 20% of a herd shows clinical signs of infection with foot and mouth disease, the daily probability that each farm type will report the occurrence of clinical signs and the number of days for which more than 20% of the herd show clinical signs Farm type Minimum time until 20% of herd shows clinical signs Daily probability of reporting No days above 20% clinical threshold Beef Dairy Sheep Pig (>500) Pig(< = 500) Beef-sheep Smallholder Feedlot/finisher 12 11 20 11 11 16 10 11 0.102 0.836 0.005 0.884 0.182 0.069 0.075 0.595 15 14 20 14 12 15 10 14 2.3 Current passive surveillance system Detection of disease through the current ‘passive’ system depends on reports of suspect clinical disease by owners or others working with livestock To model this system, the probability that a producer will report sick stock must be estimated Passive reporting by an owner relies on four steps: Clinical signs of disease must occur in the stock The owner must see/inspect the stock Detection of disease will depend on how frequently the owner observes the stock sufficiently closely to notice clinical signs The owner must recognise that the clinical signs are a problem The owner must seek help i.e report problem to veterinarian (public or private) There is no distinction between a public or private veterinarian because, in Australia, FMD is a notifiable disease and there is a legal requirement for anyone who suspects or diagnoses a disease on the National f Notifiable Diseases list to immediately notify their relevant state or territory animal health authority Under normal (pre-outbreak) conditions a small number of animals showing non-specific signs (anorexia, lameness, reduced milk production, etc) would be unlikely to trigger a suspect FMD investigation It could be expected that as the number of clinically infected animals increases so would the owner’s concern and the likelihood that they would seek assistance The actual threshold that would trigger reporting can be expected to vary with production system and individual producers Martin et al (2015) have described the development of a spreadsheet tool (General Surveillance Assessment Tool – GSAT) that has been used to assess the efficacy of general surveillance for detecting incursions of livestock diseases in Australia based on inputs from jurisdictional veterinary services Consistent with this work, we assumed a general threshold clinical prevalence of 20% in his herd for the owner to notice and seek assistance As FMD spreads at different rates in the different farm types this translates into varying minimum times to report and daily probabilities of reporting (0.005–0.884) by farm type Table summarises the daily probability of reporting by farm type, the duration of periods over which clinical signs could be reported and the expected time from first infection until the clinical threshold for possible reporting is reached Probability estimates were derived by a series of expert panels assembled to develop parameters for the GSAT, with the expected times and clinical periods derived from modelling FMD transmission within the herd (Martin et al., 2015) For each of the twelve production regions found in Australia (East et al., 2013), a separate expert panel was convened and was comprised of six to eight animal health specialists from each of the state\territory governments with land located within that M.G Garner et al / Preventive Veterinary Medicine 128 (2016) 78–86 Table The probabilities that an inspector would notice and report an FMD infected consignment (assuming ≥ 20% clinical prevalence in the source farm) Notice Problem Notify P report Abattoirs Cattle Sheep/goats Pigs 0.75 0.25 0.75 0.95 0.85 0.95 0.95 0.95 0.95 0.677 0.202 0.677 Saleyards Cattle Sheep/goats Pigs 0.5 0.3 0.5 0.95 0.55 0.95 0.95 0.95 0.95 0.451 0.157 0.451 Table The estimated probabilities and times for individual steps of the reporting chain for foot and mouth disease being reported to the chief veterinary officer (assuming ≥ 20% clinical prevalence in the source farm) Step Probability Time to occur Clinical signs occur Farmer observes stock Farmer recognises clinical signs Farmer contacts vet Vet investigates Vet suspects EAD Samples submitted to lab samples tested FMD detected CVO notified 0.7–1.0a 0.142–0.959 0.30–0.97 0.55–0.95 0.95 0.92 0.84 0.84 0.96 11–20 daysb 1–30 days day day 1–2 days – – – a Variable by animal species and by herd type, for detailed figures see Martin et al (2015) Supplementary data Table b Time from becoming infectious to reaching 20% clinical prevalence production region For the study area utilised in this paper, the Victorian Government, Department of Primary Industries also conducted a producer survey to inform the probability estimates used in this study (results not published) In addition to on-farm reporting by the owner/manager, suspicion of FMD associated with clinical signs could also be reported by an abattoir or saleyard inspector as part of their normal inspection role Probabilities that an inspector would notice, recognise and report an FMD infected consignment (Table 3) were derived by a series of expert panels assembled to develop parameters for the GSAT (Martin et al., 2015) Tables and provide the basis for determining whether suspect FMD will be recognised and investigated However, several additional steps are necessary to ensure that FMD is confirmed and reported to the CVO These additional steps are listed in Table Again, these estimates for these steps were derived by a series of expert panels assembled to develop parameters for the GSAT and were assumed to be constant for all farm types and geographic regions (Martin et al., 2015) Estimates of the time elapsed at each step of the passive surveillance chain were derived from a review of the literature and by the series of expert panels assembled to develop parameters for the GSAT (Martin et al., 2015) This expected time required for investigation and notification of the CVO will vary depending on the nature of the report and the following values have been used: Farmer report – days, Saleyard inspection – days and Abattoir inspection – days The final values used are shown in Table In Australia, if a notifiable disease like FMD is suspected, jurisdictional animal health services have the legal instruments to quarantine the suspected premises Australia’s policy is to also implement a national livestock standstill from the time of diagnosis of FMD or on strong suspicion of the disease (Animal Health Australia, 2015) 81 2.4 Enhancing the ‘passive’ surveillance system The reporting of suspect disease by livestock owners could be improved by increasing their awareness of serious disease, by increasing the frequency and extent of livestock inspections by owners, and by improving willingness of owners to have problems investigated (e.g by a government or private veterinarian) Education and awareness programs could be generalised or targeted to specific industry sectors From a modelling perspective, this can be represented by a reduced clinical threshold for reporting to occur and/or an increased daily probability that a livestock owner with clinically affected livestock would report However, it is difficult to predict the impact of awareness programs on producer behaviour A review of extension and producer education programs conducted in Australia in recent years suggests that the best programs result in an uptake of new farming practices of only 30–40% of participants (Larsen et al., 2002; Price and Hacker, 2009; Hunt et al., 2011) Although it cannot be assumed that producers would respond in the same way to biosecurity awareness, in the absence of other information we believe that this provides a plausible upper estimate of what might be expected Swill feeding to pigs is the most likely opportunity for FMD to establish in Australia (Matthews, 2011), we evaluated an education campaign targeted at the small-to-medium pig producers to which 30% of producers responded; the daily probability of reporting by a pig producer with a herd showing clinical signs consistent with FMD increased from a baseline of 0.18 to 0.48 2.5 Active surveillance at saleyards Active surveillance programs for diseases like FMD could be based on testing samples of animals submitted for sale This testing could be based on random or targeted sampling regimes In the former case animals would be selected at random from the population on any given sale day In the latter case, animals would be selected for sampling on the basis of some form of screening procedure e.g animals showing signs like lameness or through infra-red thermography (Rainwater-Lovett et al., 2009) The probability that a single infected farm submitting animals to a sale would be detected was estimated, based on simple random sampling, assuming 20, 50 or 100 animals are tested per sale and FMD test sensitivity of 97% (King et al., 2006) The probability that a single infected farm would be detected was also estimated assuming the saleyard population is first screened with infrared thermography, which was assumed to have a sensitivity of 61.1% and a specificity of 87.7% (RainwaterLovett et al., 2009) 2.6 Bulk milk testing Although bulk milk testing shows good potential as a screening test for FMD, there is no commercially available test at this time For this study, we made the following assumptions: • all dairy farms will be tested every day (milk is collected daily from dairy farms in the study region) A result is available the next day after the milk is collected • testing is at the tanker level • the PCR has an analytical sensitivity of 10 −2.5 or 10 −3 With an average milking herd size of 225, this means four or two infected cows per herd, assuming a minimum of one infected herd contributing to a tanker Our modelling used three infected cows as the farm level threshold for detection (∼10−2.6 ) • the diagnostic sensitivity of the milk PCR test is 95% • there is a two day delay from when milk is tested until FMD is confirmed – to allow for traceback of individual farms and confirmatory investigation/testing 82 M.G Garner et al / Preventive Veterinary Medicine 128 (2016) 78–86 2.7 Outbreak scenarios The source of the FMD introduction was assumed to be contaminated food brought into Australia by a traveller returning from overseas and then (illegally) fed to pigs This is considered to be the most likely way for FMD to establish in Australia (Matthews, 2011) This was assumed to occur on a smaller pig farm since in Australia larger pig farms typically have high levels of biosecurity (Hernández-Jover et al., 2012) The introduction was assumed to occur in May, when cold and wet conditions would favour FMD virus survival outside a host For each model iteration, a pig farm of the appropriate type was randomly selected as the source farm to initiate an outbreak The distribution of smaller pig farms in the study area is shown in Fig 2.8 Study design AusSpread was configured so that the first detection of FMD was dynamically determined, taking into account the above detection parameters Disease was introduced into a randomly selected small pig farm ( 0.05) (Table 7) There was no difference in the efficacy of bulk milk testing whether sampling was conducted daily, every two days or every three days (Table 7) Deploying an integrated surveillance system where bulk milk testing only was used in the dairying areas and producer education 84 M.G Garner et al / Preventive Veterinary Medicine 128 (2016) 78–86 Table The delay to detection, duration and size of foot and mouth disease outbreaks [median (95% probability interval)] in Victoria, Australia using different surveillance systems Surveillance Delay to detection (days) Duration(days) Total farms culled Baseline passive Saleyard 20 BMT1 BMT2 BMT3 integrated 20 (16–46)a, * 20 (16–42)a 18 (9–40)c 18 (9–40)c 18 (9–40)c 17 (12–24)b,c 53 (34–106)a 53 (34–99)a 46 (14–78)a,b 46 (14–76)a,b 47 (14–76)a,b 45 (32–76)b 46 (5–363)a 46 (5–310)a 40 (5–183)c 43 (5–181)c 40 (5–178)c 37 (6–176)b,c * Within each column, figures with the same superscript are not significantly different to enhance reporting by pig farmers in other areas did not provide a better outcome (p > 0.05) than either bulk milk testing alone or producer education alone (Table 7) Discussion Early detection of disease incursions is an objective of all veterinary services because it results in smaller outbreaks that are easier to control and eradicate (Carpenter et al., 2011) Achieving earlier detection also has significant economic benefits because the costs associated with disease control will be less and loss of export earnings due to exclusion from FMD sensitive markets will also be reduced Worldwide, recent history suggests that outbreaks of FMD in previously free countries are detected about three weeks after introduction of the virus to the index farm (Anderson, 2002; Bouma et al., 2003; Yoon et al., 2013); this is consistent with the modelling results for the current Australian surveillance system presented in this paper We have shown that even with 100% reporting by producers in the study area, detection of FMD through passive surveillance cannot be realistically expected earlier than 15 days after introduction to the index farm because of the time delays associated with: the incubation period of the disease; spread of the disease to a prevalence within a herd to trigger action by the owner, veterinary investigation, collection of samples and laboratory testing In our study region, the difference in the delay to detection between the current surveillance system and a gold standard system is only five days so there is only a small window for improvement Despite this, the associated reduction in the duration and extent of the outbreak could be associated with significant economic benefits In this paper, we have concentrated on the most likely pathway of entry of FMD into Australia i.e swill feeding of FMD contaminated material to pigs Whilst other pathways of entry exist, they are considered less likely (Matthews, 2011; East et al., 2013) The only other significant pathway of introduction into Australia viz fomites may lead to the initial introduction occurring on a different type of farm however, once the FMD spreads to a second and subsequent farms, the pattern of spread and likelihoods of detection are identical Therefore, substantial changes in the observed results are unlikely Improvements in the time to detection could be made, if disease could be detected in the pre-clinical phase The specific method of achieving early detection must be practical and cost effective Implementation of active surveillance at saleyards, whilst theoretically effective, would require collecting and processing a large number of samples (∼700 per sale) for PCR testing in real time Even pre-screening the sale population with infrared thermography, assuming an average sized sale of 1350 animals, at one animal per minute would take 22.5 h Australia has international airports at its geographic extremities and preserved meat smallgoods and dairy produce are routinely seized by quarantine authorities at each of these airports (confidential department reports) Further, incoming passengers routinely disperse widely and rapidly from their point of arrival to their final destination Our previous publication (East et al., 2013) recognised this and used only the location of pig farms as the most likely point of entry of FMD because any FMD contaminated goods need to be fed to a pig to introduce FMD Further, it is not unusual in Australia (East and Foreman, 2011; East et al., 2014a,b) for animals to be moved hundreds (occasionally thousands) of kilometres to a saleyard Therefore, any program would have to function at all saleyards to be effective The program would also need to be on-going because we cannot predict when FMD will enter Australia In addition, testing large numbers of animals even with tests with high specificity will inevitably result in false positive results that would cause major logistical issues for saleyard operators and animal health authorities as the facility would have to be locked down until the situation was resolved A similar conclusion was reached with bulk milk testing For early detection, BMT testing would also need to be widespread and ongoing FMD is shed in milk during the incubation period In experimental studies, Reid et al (2006) found that the earliest period for detection of FMD in milk was 2–3 days post inoculation and that milk collected from an infected cow and diluted 10,000 fold was able to be detected using the rRT-PCR tests The average size of an Australian dairy herd is 225 cows and the average yield is 17 l/cow/day This means that the milk from four to five farms (900–1125 cows) would be combined into a 20,000 l milk tanker In the early stages of infection, very few animals in a herd are likely to be infected but if PCR can return a positive result after milk has been diluted 10000 fold, testing at the tanker level could allow detection of one infected cow The results obtained with bulk milk testing were very dependent on where the outbreak starts; larger outbreaks occurred in the more intensive areas The median size of outbreaks size when an incursion started in a dairy area was twice the size of outbreaks starting in non-dairy areas Whilst bulk milk testing reduced the time to detection by three days compared to the Baseline, when the outbreak commenced in a dairy area, the reduction was only one day when the outbreak started outside the dairy area (Results not shown) Whilst these reductions were suggestive of enhanced early detection, the differences were not statistically significant In contrast to our study, Thurmond and Perez (2006) found that bulk milk testing provided an effective method of enhancing early detection of FMD infection This difference can be explained because Thurmond and Perez were examining the efficacy of bulk milk testing for detection of FMD within an isolated, infected dairy herd The design of our study specifically requires that infection first occurs within a small-to-medium pig herd because that is the most likely pathway of entry of FMD into Australia (Matthews, 2011) In our scenario therefore, the infection needs to spread to a dairy farm before it can be detected by bulk-milk screening The delay to infection of the first dairy farm is a function of the geographic location of the initially infected pig farm and the stochastic probability of spread of FMD to other farms This time delay to infection of the first dairy farm cancels out any advantage of early detection provided by bulk-milk testing M.G Garner et al / Preventive Veterinary Medicine 128 (2016) 78–86 Daily testing of 30–50 samples per day for each of the 25 milk factories in the study area would require testing of 450,000 samples per year Although logistically this could be done, it would be expensive In a separate study we report an economic analysis of circumstances under which BMT is cost effective (Kompas et al submitted) As with active surveillance at saleyards, test specificity issues would be a concern To investigate a positive result, all individual farms contributing to the bulk milk consignment would need to be followed up By the time follow-up testing was completed, it is likely that clinical signs would have occurred on the affected farm, negating much of the time advantage from the original test While BMT shows considerable promise as an FMD surveillance tool, it appears to be better suited to surveillance during an outbreak and for post-outbreak proof of freedom testing (Kompas et al submitted) We conclude that, of the possibilities examined in this paper, enhancing the passive surveillance system offers the best opportunity for early detection of FMD in Victoria This will have particular benefits in reducing the likelihood of getting a large outbreak Maintaining a high level of awareness of FMD and promoting the need for disease reporting by all producers and those involved in livestock production is essential If resources are limited, targeting education programs to those sectors that are considered to be at greater risk of being a source of introduction or where current disease reporting and investigation is low is likely to be most effective Acknowledgment This work was funded in part by the Centre of Excellence for Biosecurity Risk Analysis References ABARES, 2014 Agricultural Commodity Statistics 2014 Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra, December Animal Health Australia, 2014a Disease strategy: foot-and-mouth disease (version 3.4) In: Australian Veterinary Emergency Plan (AUSVETPLAN), third ed Standing Council on Primary Industries, Canberra, ACT Animal Health Australia, 2015 Animal Health in Australia 2014 http://www animalhealthaustralia.com.au/ (accessed 7.10.15) Animal Health Australia, 2014b AUSVETPLAN Guidance Document: Declared Areas and Allocation of Premises Classifications in an EAD Response, second ed, accessed 20.09.15 http://www.animalhealthaustralia.com.au/programs/ emergency-animal-disease-preparedness/ausvetplan/guidance-documents/ Anderson, I., 2002 Foot and Mouth Disease 2001: Lessons to Be Learned Inquiry Report.p 21, accessed 08.01.15 http://wildpro.twycrosszoo.org/S/00Ref/ BooksContents/B494 LessonsLearned/B494 LessonsLearned.htm Bates, T.W., Thurmond, M.C., Hietala, S.K., Venkateswaran, K.S., Wilson, T.M., Colston, J.E., Trebes Jr., B.W., Milanovich, F.P., 2003 Surveillance for detection of foot-and-mouth disease J Am Vet Med Assoc 223, 609–614 Blackwell, J.H., McKercher, P.D., Kosikowski, F.V., Carmichael, L.E., Gorewit, R.C., 1982 Concentration of foot-and-mouth disease virus in milk of cows infected under simulated field conditions J Dairy Sci 65, 1624–1631 Blayney, D.P., Dyck, J., Harvey, D., 2006 Economic effects of animal diseases linked to trade dependency Amber Waves 4, 23–29 Bouma, A., Elbers, A.R.W., Dekker, A., de Koeijer, A., Bartels, C., Vellema, P., van der Wal, P., van Rooij, E.M.A., Pluimers, F.H., de Jong, M.C.M., 2003 The foot-and-mouth disease epidemic in The Netherlands in 2001 Prev Vet Med 57, 155–166 Carpenter, T.E., O’Brien, J.M., Hagerman, A.D., McCarl, B.A., 2011 Epidemic and economic impacts of delayed detection of foot-and-mouth disease: a case study of a simulated outbreak in California J Vet Diagn Invest 23, 26–33 East, I.J., Foreman, I., 2011 The structure, dynamics and movement patterns of the Australian sheep industry Aust Vet J 89, 477–489 East, I.J., Wicks, R.M., Martin, P.A.J., Sergeant, E.S.G., Randall, L.A., Garner, M.G., 2013 Use of a multi-criteria analysis framework to inform the design of risk based general surveillance systems for animal disease in Australia Prev Vet Med 112, 230–247 East, I.J., Roche, S.E., Wicks, R.M., de Witte, K., Garner, M.G., 2014a Options for managing animal welfare on intensive pig farms confined by movement restrictions during an outbreak offoot and mouth disease Prev Vet Med 117, 533–541 East, I.J., Davis, J., Sergeant, E.S.G., Garner, M.G., 2014b The structure, dynamics and movement patterns of the Australian pig industry Aust Vet J 92, 52–57 85 East, I.J., Martin, P.A.J., Langstaff, I., Iglesias, R.M., Sergeant, E.S.G., Garner, M.G., 2016 Assessing the delay to detection and the size of the outbreak at the time of detection of incursions of foot and mouth disease in Australia Prev Vet Med 123, 1–11 Elbers, A.R.W., Stegeman, J.A., Moser, H., Ekker, H.M., Smak, J.A., Pluimers, F.H., 1999 The classical swine fever epidemic 1997–1998 in The Netherlands: descriptive epidemiology Prev Vet Med 42, 157–184 Elbers, A.R.W., Gorgievski-Duijvesteijn, M.J., van der Velden, P.G., Loeffen, W.L.A., Zarafshani, K., 2010 A socio-psychological investigation into limitations and incentives concerning reporting a clinically suspect situation aimed at improving early detection of classical swine fever outbreaks Vet Microbiol 142, 108–118 Frawley, P.T., 2003 Review of Rural Veterinary Services Department of Agriculture, Fisheries and Forestry, Canberra (accessed 08.01.15) https://www ava.com.au/sites/default/files/documents/Other/Frawley%20report.pdf Garner, M.G., Beckett, S.D., 2005 Modelling the spread of foot and mouth disease in Australia Aust Vet J 83, 30–38 Garner, M.G., Bombarderi, N., Cozens, M., Conway, M.-L., Wright, T., Paskin, R., East, I.J., 2014 Estimating resource requirements to staff a response to a medium to large outbreak of foot-and-mouth disease in Australia Transbound Emerg Dis., http://dx.doi.org/10.1111/tbed.12239 Grubman, M.J., Baxt, B., 2004 Foot and mouth disease Clin Microbiol Rev 17, 465–493 Hernández-Jover, M., Cogger, N., Martin, P.A.J., et al., 2011 Evaluation of post-farm-gate passive surveillance in swine for the detection of foot and mouth disease in Australia Prev Vet Med 100, 171–186 Hernández-Jover, M., Gilmour, J., Schembri, N., Sysak, T., Holyoake, P.K., Beilin, R., Toribio, J.-A.L.M.L., 2012 Use of stakeholder analysis to inform risk communication and extension strategies for improved biosecurity amongst small-scale pig producers Prev Vet Med 104, 258–270 Hoinville, L.J., 2011 Animal Health Surveillance Terminology Final Report from Pre-ICAHS Workshop (accessed 19.01.15) http://webarchive.nationalarchives gov.uk/20140707135733/http://www.defra.gov.uk/ahvla-en/files/icahsworkshop-report.pdf Hopp, P., Vatn, S., Jarp, J., 2007 Norwegian farmers’ vigilance in reporting sheep showing scrapie-associated signs BMC Vet Res., http://dx.doi.org/10.1186/ 1746-6148-3-34 Hunt, W., Vanclay, F., Birch, C., Coutts, J., Flittner, N., Williams, B., 2011 Agricultural extension: building capacity and resilience in rural industries and communities Rural Soc 20, 112–127 King, D.P., Ferris, N.P., Shaw, A.E., Reid, S.M., Hutchings, G.H., Giuffre, A.C., Robida, J.M., Callahan, J.D., Nelson, W.M., Beckham, T.R., 2006 Detection of foot-and-mouth disease virus: comparative diagnostic sensitivity of two independent real-time reverse transcription-polymerase chain reaction assays J Vet Diagn Invest 18, 93–97 Klinkenberg, D., Nielen, M., Mourits, M.C.M., de Jong, M.C.M., 2005 The effectiveness of classical swine fever surveillance programs in The Netherlands Prev Vet Med 67, 19–37 Knight-Jones, T., Rushton, J., 2013 The economic impacts of foot and mouth disease What are they, how big are they and where they occur? Prev Vet Med 112, 161–173 Kompas, T., Nguyen, H.T.M., Ha, P.V., 2015 Food and biosecurity: livestock production and towards a world free of foot-and-mouth disease Food Secur 7, 291–302 Langstaff, I., 2008 National animal health strategy reference group meeting Anim Health Surveill Q Rep 13, 5–6 Larsen, J., Vizard, A., Counsell, D., Scrivener, C., Hygate, L., Webb Ware, J., 2002 Linking Australian woolgrowers with research: the South Roxby project Wool Technol Sheep Breed 50, 266–273 Mansley, L.M., Dunlop, P.J., Whiteside, S.M., Smith, R.G., 2003 Early dissemination of foot-and-mouth disease virus through sheep marketing in February 2001 Vet Rec 153, 43–50 Martin, P.A.J., Langstaff, I., Iglesias, R.M., East, I.J., Sergeant, E.S.G., Garner, M.G., 2015 Assessing the efficacy of general surveillance for detection of incursions of livestock diseases in Australia Prev Vet Med 121, 215–230 Matthews, K., 2011 A Review of Australia’s Preparedness for the Threat of Foot-and-mouth Disease Australian Government, Department of Agriculture, Fisheries and Forestry, Canberra (accessed 08.01.15) http://www.daff.gov.au/ animal-plant-health/pests-diseases-weeds/animal/fmd/review-foot-andmouth-disease Nairn, M.E., Allen, P.G., Inglis, A.R., Tanner, C., 1996 Australian Quarantine: a Shared Responsibility Department of Primary Industries and Energy, Canberra (accessed 08.01.15) http://www.agriculture.gov.au/biosecurity/about/reportspubs/nairn OIE, 2012 The Global Foot and Mouth Disease Control Strategy: Strengthening Animal Health Systems Through Improved Control of Major Diseases (accessed 05.05.15) http://www.oie.int/doc/ged/D11886.PDF Palmer, S., Fozdar, F., Sully, M., 2009 The effect of trust on Western Australian Farmers’ responses to infectious livestock diseases Sociol Ruralis 49, 360–374 Price, R.J., Hacker, R.B., 2009 Grain & Graze: an innovative triple bottom line approach to collaborative an multidisciplinary mixed-farming systems research, development and extension Anim Prod Sci 49, 729–735 86 M.G Garner et al / Preventive Veterinary Medicine 128 (2016) 78–86 Rainwater-Lovett, K., Pacheco, J.M., Packer, C., Rodriguez, L.L., 2009 Detection of foot-and-mouth disease virus infected cattle using infrared thermography Vet J 180, 317–324 Reid, S.M., Parida, S., King, D.P., Hutchings, G.H., Shaw, A.E., Ferris, N.P., Zhang, Z., Hillerton, E., Paton, D.J., 2006 Utility of automated real-time RT-PCR for the detection of foot-and-mouth disease virus excreted in milk Vet Res 37, 121–132 Thurmond, M.C., Perez, A.M., 2006 DVM, Modeled detection time for surveillance for foot-and-mouth disease virus in bulk tank milk Am J Vet Res 67, 2017–2024 Yoon, H., Yoon, S.-S., Kim, Y.-J., Moon, O.-K., Wee, S.-H., Joo, Y.-S., Kim, B., 2013 Epidemiology of the foot-and-mouth disease serotype O epidemic of November 2010 to April 2011 in the Republic of Korea Transbound Emerg Dis., http://dx.doi.org/10.1111/tbed.12109  ... delay to detection, duration and size of foot and mouth disease outbreaks [median (95% probability interval)] in Victoria, Australia using different surveillance systems Surveillance Delay to detection... therefore of interest to examine whether potential exists to improve time to detection through introducing active surveillance, using new methods of surveillance or enhancing the existing passive surveillance. .. probability interval)] of the largest quartile of foot and mouth disease outbreaks in Australia using different surveillance systems Surveillance System Baseline Enhanced Passive Passive * Delay to detection