Preventive Veterinary Medicine 81 (2007) 3–20 www.elsevier.com/locate/prevetmed

The application of risk analysis in aquatic animal health management E.J. Peeler a,*, A.G. Murray b, A. Thebault c, E. Brun d, A. Giovaninni e, M.A. Thrush f a

EpiCentre, Private Bag 11222, Institute of Veterinary and Animal Biomedical Sciences, Massey University, Palmerston North, New Zealand b Fisheries Research Services, Marine Laboratory, PO Box 101, 375 Victoria Road, Aberdeen AB11 9DB, UK c Agence Franc¸ais de Securite Sanitaire des Aliments, Paris, France d National Veterinary Institute, PO Box 8156, 0033 Oslo, Norway e Istituto Zooprofilattico Spermentale dell’Abruzzo e del Molise, Via Campo Boario 64100, Teramo, Italy f Centre for Environment, Fisheries and Aquaculture Science, Barrack Road, Weymouth DT4 8UB, UK

Abstract Risk analysis has only been regularly used in the management of aquatic animal health in recent years. The Agreement on the Application of Sanitary and Phytosanitary measures (SPS) stimulated the application of risk analysis to investigate disease risks associated with international trade (import risk analysis—IRA). A majority (9 of 17) of the risk analyses reviewed were IRA. The other major focus has been the parasite of Atlantic salmon—Gyrodactylus salaris. Six studies investigated the spread of this parasite, between countries, rivers and from farmed to wild stocks, and clearly demonstrated that risk analysis can support aquatic animal health policy development, from international trade and biosecurity to disease interaction between wild and farmed stocks. Other applications of risk analysis included the spread of vertically transmitted pathogens and disease emergence in aquaculture. The Covello–Merkhofer, risk analysis model was most commonly used and appears to be a flexible tool not only for IRA but also the investigation of disease spread in other contexts. The limitations of the identified risk assessments were discussed. A majority were * Corresponding author. Present address: Cefas, Barrack Rd, Weymouth, DT4 8UB, UK. Tel.: +44 1305 206746; fax: +44 1305 206601. E-mail address: [email protected] (E.J. Peeler). 0167-5877/$ – see front matter # 2007 Published by Elsevier B.V. doi:10.1016/j.prevetmed.2007.04.012

4

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

qualitative, partly due to the lack of data for quantitative analysis, and this, it can be argued, constrained their usefulness for trade purposes (i.e. setting appropriate sanitary measures); in other instances, a qualitative result was found to be adequate for decision making. A lack of information about the disease hazards of the large number of fish species traded is likely to constrain quantitative analysis for a number of years. The consequence assessment element of a risk analysis was most likely to be omitted, or limited in scope and depth, rarely extending beyond examining the evidence of susceptibility of farmed and wild species to the identified hazard. The reasons for this are discussed and recommendations made to develop guidelines for a consistent, systematic and multi-disciplinary approach to consequence assessment. Risk analysis has improved decision making in aquatic animal health management by providing a transparent method for using the available scientific information. The lack of data is the main constraint to the application of risk analysis in aquatic animal health. The identification of critical parameters is an important output from risk analysis models which should be used to prioritise research. # 2007 Published by Elsevier B.V. Keywords: Risk analysis; Epidemiology; Aquatic; Fish; Health

1. Introduction Veterinary epidemiology can be defined as the study of disease and its determinants in populations to improve animal health and welfare (Dohoo et al., 2003). To achieve its purpose, veterinary epidemiology draws upon and integrates methods from a number of disciplines including economics, statistics, animal production and risk analysis. Risk analysis investigates both the likelihood and consequences of undesirable events, known as hazards (Vose, 2000). In the field of animal health, risk analysis has been used mainly in the assessment of disease introduction with international trade, known as import risk analysis (IRA). The development of IRA has been driven by the Agreement on the Application of Sanitary and Phytosanitary (SPS) Measures (WTO, 1995) of the World Trade Organisation (WTO), which requires an IRA to justify sanitary measures over and above those sanctioned by international agreement. Guidelines for IRA (O.I.E., 2004) stipulate that the data and assumptions used in the analysis are comprehensively documented, thereby improving the rigour and transparency of decision making in the face of uncertainty. In this paper the use of risk analysis in aquatic animal health management is reviewed, the potential for future applications and methodological developments are considered. Risk assessments of environmental stressors, e.g. contaminants, on the health of aquatic ecosystems have used fish health as one of many indicators (Adams et al., 2000). Since these studies do not support aquatic animal health management, they are not considered in this review.

2. Risk assessment and aquatic animal health—a historical perspective The key events in the application of risk assessment in aquatic animal health are set out in Table 1. Governments have for many years sought to prevent the introduction of alien diseases and animal and plant species through sanitary measures. In general countries based sanitary measures on expert opinion and literature reviews. In the UK the Diseases of Fish act, enacted in 1937 (Hill, 1996), is an early example of legislation to protect the

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

5

Table 1 Key events in the application of risk assessment in aquatic animal health Date

Description

1993

Publication by OIE of a Scientific and technical review on ‘Risk analysis, animal health and trade’ (O.I.E., 1993), which included an import risk assessment for salmon meat (Beers and Wilson, 1993) Adoption of the Agreement on the Application of Sanitary and Phytosanitary measures by the World Trade Organisation (which required members to justify sanitary barriers, not sanctioned by international agreements, with a science-based risk analysis) (WTO, 1995) Inclusion of guidelines on import risk assessment in the OIE Aquatic Animal Health Code WTO appellate body finding on the ‘Australia—measures affecting importation of salmon’ case that the Australian measure was not based on a proper risk assessment and that imports of other fish presented similar or greater risk (Trachtman, 1999) The use of quantitative risk assessment to inform the management of the fish parasite, Gyrodactylus salaris in Norway (Paisley et al., 1999)—early use of quantitative risk analysis to support fish health management decision making to protect wild stocks OIE conference on ‘Risk analysis in aquatic animal health’ (O.I.E., 2001)

1995

1995 1998

1999

2000

health status of fish. The ‘Furunculosis committee’ (Mackie et al., 1935) concluded that the disease had been introduced to the UK with the importation of trout leading to a ban on the imports of live salmonids in the 1937 legislation. Thus the UK adopted a ‘zero-risk’ approach to the introduction of exotic disease through trade. The threat of disease introduction has been used in some instances to enforce stringent national measures (e.g. import bans) designed primarily to obstruct trade and protect local industries (Zepeda et al., 2001). In recognition of this problem, article 20 of the General Agreement on Tariffs and Trade (GATT) allowed governments to restrict trade in order to protect human, animal or plant life or health, provided they do not discriminate or use sanitary measures as disguised protectionism. In 1995 the WTO replaced the GATT. The agreement on the application of Sanitary and Phytosanitary measures (the SPS agreement) (WTO, 1995) came into force with the establishment of the WTO. The SPS agreement sought to address the perceived failure of the GATT measures to prevent the use of sanitary measures as a form of protectionism (Roberts et al., 1999). Its main purpose is to facilitate trade whilst allowing protection of human, animal and plant life (recognising that trade cannot be risk free). The SPS agreement recognised the OIE as the body responsible for setting animal health standards. The SPS agreement gave countries the right to protect the human, animal or plant life or health provided they did not contravene the articles of the agreement. Members are required to produce a scientific justification for measures that produce a higher level of protection than provided for under international agreements or guidelines. The SPS agreement indicates that Members, when developing sanitary measures, take into account risk assessment methods developed by the OIE (the recognised standard setting body). Thus the SPS agreement sought to make the setting of sanitary measures science-based and recommended risk assessment as a suitable method. This can be viewed as a formalisation of the previous approach, which relied on expert opinion. Risk analysis should provide a scientific evaluation of the likelihood and consequences of the identified hazards, thus narrowing the information gap between importers and exporters and enabling common judgements about level of risk mitigating measures (Roberts et al., 1999).

6

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

Once established through its application in international trade, the use of risk analysis extended into other areas of aquatic animal health; an early example of which is work by Paisley et al. (1999) who investigated the risk of the spread of a parasite, Gyrodactylus salaris, from farmed to wild Atlantic salmon (Salmo salar).

3. Approaches 3.1. Risk assessment models The three commonly used risk assessment models are: the US National Academy of Sciences and National Research Council (NAS–NRC) model, the Covello–Merkhofer (CM) model and the International Plant Protection Convention (IPPC) model. These models share many common features, most notably, the use of scenario trees to describe the temporal relationship between the series of events necessary for the hazard to occur. The guidelines for IRA published in the Aquatic Animal Health Code (O.I.E., 2004) of the World Organisation for Animal Health (O.I.E.) are based on the CM model (Covello and Merkhofer, 1993). These guidelines have been generally preferred for aquatic animal risk analysis. There are four parts to the CM risk analysis model (Fig. 1). Hazard identification is the first step and considered separately from the risk assessment. Risk assessment is subdivided into stages: (i) release assessment (pathways for the introduction of the hazard), (ii) exposure assessment (pathways necessary for the hazard to occur following introduction), (iii) consequence assessment (identification of the adverse human health, animal health, economic or environmental effects), and (iv) risk estimation (integration of the release, exposure and consequence assessments). Risk management and risk communication combined with risk assessment and hazard identification are referred to as ‘‘risk analysis’’. The CM model can be applied qualitatively or quantitatively. 3.2. Qualitative and quantitative approaches Risk analyses are generally described as qualitative or quantitative. A qualitative assessment is essentially a reasoned, systematic and logical discussion of the relevant contributory factors and epidemiology of a hazard, in which the likelihood of its release and

Fig. 1. The Covello–Merkhofer risk analysis model (Covello and Merkhofer, 1993).

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

7

exposure and the magnitude of its consequences are expressed using non-numerical terms such as high, medium, low or negligible (Murray et al., 2004). In quantitative assessments, the likelihood of the release and exposure of the hazard are expressed numerically. The uncertainty associated with an input, and its known variability, can be modelled by probability distributions. Monte Carlo simulation, using software programmes (e.g. @RISK, Palisade Inc.), is generally used in quantitative risk analyses to assimilate the probability components, and allows sensitivity analysis to identify parameters to which the outcome is most sensitive. The choice between adopting a qualitative or quantitative approach is determined by a number of factors. Generally it is recommended that all risk analyses should first be attempted qualitatively (Vose, 2001). Expending further resources on a quantitative assessment depends, firstly, on whether the qualitative results were adequate for decision making and, secondly, on whether resources and data are available for a quantitative analysis. Qualitative approaches have been mainly used in aquatic animal health, in some cases, because the outcome may have been adequate to support decision making, but also because the lack of data has prevented a quantitative approach (Rodgers, 1997).

4. Disease spread A number of key word searches of abstracting databases (Aquatic Sciences and Fisheries Abstracts and Web of Science) and Internet search engines (Google and Google Scholar) and the knowledge of the authors were used to identify 17 risk analyses for the spread of aquatic animal diseases (Table 2). Only articles written in English were sought. It is likely that a number of risk analyses will not have been located by the methods used; this review, therefore, cannot be considered as comprehensive. The papers were categorised by the approach (qualitative or quantitative) and scope (spread between countries, regions, rivers or farms) (Table 2); the species, commodities and pathogens investigated (Table 3) and the components of the risk analysis process completed (Table 4). Eight studies investigated disease spread between countries (i.e. IRA), three investigated spread between regions of the same country, four studied disease spread between rivers and two investigated spread between farms. Both qualitative (10 studies) and quantitative (6 studies) approaches had been adopted. Nine papers considered single pathogens, seven considered multiple pathogens and one was a generic model for vertically transmitted pathogens (Peeler et al., 2005). With the exception of Bondad-Reantaso et al. (2005), the studies were undertaken in developed economies (i.e. Australia, New Zealand, North America and Europe). The geographical distribution is reflected in the species studied (12 reports investigated salmonid diseases). All but three studies (Mortensen, 2000; Edgerton, 2002; Bondad-Reantaso et al., 2005) investigated finfish and their diseases (Table 3). Six studies investigated the spread of G. salaris, a reflection of the threat that this parasite represents to wild stocks of Atlantic salmon. 4.1. Import risk analyses The IRA identified range from papers given at a scientific conference (Bruneau, 2001; Pharo and MacDiarmid, 2001), papers in peer-reviewed journals (Peeler and Thrush, 2004)

Import risk analysis on non-viable salmonids and non-salmonids marine finfish Import risk analysis on live ornamental finfish Qualitative analysis of the risk of introducing G. salaris into the United Kingdom Hazard analysis of exotic pathogens of potential threat to European freshwater crayfish Pathogen and ecological risk analysis for the introduction of blue shrimp, Litopenaeus stylirostris, from Brunei Darussalam to Fiji An assessment of the risk of spreading the fish parasite G. salaris to uninfected territories in the European Union with the movement of live Atlantic salmon (Salmo salar) from coastal waters The risk of introducing exotic diseases of fish into New Zealand through the importation of ocean-caught Pacific salmon from Canada Import health risk analysis: salmonids for human consumption Scallop introductions and transfers, from an animal health point of view A quantitative risk assessment for the introduction of Myxobolus cerebralis to Alberta, Canada, through the importation of live salmonids. Risk analysis in aquatic animal health Negligible risk associated with the movement of processed rainbow trout, Oncorhynchus mykiss (Walbaum), from an infectious haematopoietic necrosis virus (IHNV) endemic area Risk of inter-river transmission of G. salaris by migrating Atlantic salmon smolts, estimated by Monte Carlo simulation An evaluation of potential routes for spreading G. salaris A Monte Carlo simulation model for assessing the risk of introduction of G. salaris to the Tana river, Norway Qualitative risk assessment of routes of transmission of the exotic fish parasite G. salaris between river catchments in England and Wales The application of risk assessment to the study of vertical transmission of fish pathogens An evaluation of the relative risks of infectious salmon anaemia transmission associated with different salmon harvesting methods in Scotland

Kahn et al. (1999a) Kahn et al. (1999b) Peeler and Thrush (2004) Edgerton (2002) Bondad-Reantaso et al. (2005) Peeler et al. (2006)

Peeler et al. (2005) Munro et al. (2003)

Peeler et al. (2004)

Jansen et al. (2006) Paisley et al. (1999)

Høga˚sen and Brun (2003)

LaPatra et al. (2001)

Stone et al. (2001) Mortensen (2000) Bruneau (2001)

MacDiarmid (1994)

Title

Authors

Table 2 Summary of aquatic animal risk analyses for disease spread: approach and scope

Qualitative Qualitative

Qualitative

Quantitative Quantitative

Quantitative

Qualitative

Quantitative Qualitative Quantitative

Quantitative

Trans-farm Trans-farm

Trans-river

Trans-river Trans-river

Trans-river

Trans-region

Trans-country Trans-region Trans-region

Trans-country

Trans-country

Qualitative

Scope Trans-country Trans-country Trans-country Trans-country Trans-country

Approach Qualitative Qualitative Qualitative Qualitative Qualitative

8 E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

9

Table 3 Summary of disease spread aquatic animal risk analyses: species, commodity and pathogens Authors

Susceptible species

Commodity/route

Pathogen

Kahn et al. (1999a)

Finfish (multi spp.)

All

Kahn et al. (1999b) Peeler and Thrush (2004) Edgerton (2002) Bondad-Reantaso et al. (2005) Peeler et al. (2006) MacDiarmid (1994) Stone et al. (2001) Mortensen (2000) Bruneau (2001) LaPatra et al. (2001)

Ornamental finfish Salmonids Crayfish spp. Blue shrimp Atlantic salmon Salmonids Salmonids Scallops Rainbow trout Rainbow trout

Høga˚sen and Brun (2003) Jansen et al. (2006) Paisley et al. (1999) Peeler et al. (2004) Peeler et al. (2005) Munro et al. (2003)

Atlantic salmon Atlantic salmon Atlantic salmon Salmonids Salmonids Salmonids

Non-viable whole fish and fish products (for human consumption) Live fish Multiple routes Live crayfish Live crustacea Live fish Fish products Fish products Live shellfish Live fish Non-viable whole fish and fish products (for human consumption) Live fish (migration) Multiple routes Live fish Multiple routes Sale of fish eggs Multiple routes

All G. salaris All All G. salaris All All All M. cerebralis IHNV

G. salaris G. salaris G. salaris G. salaris Generic Infectious salmon anaemia

All = all economically and environmentally important fish pathogens associated with the commodity or route; generic: risk model not specific to an individual pathogen.

and reports by government departments (MacDiarmid, 1994; Kahn et al., 1999a,b; Stone et al., 2001) or international agencies (Bondad-Reantaso et al., 2005). A number of the IRA were produced by the regulatory authorities in Australia and New Zealand. IRA completed by the Australian authorities include studies of salmon and other marine fish from Canada (Kahn et al., 1999a), ornamental fish (Kahn et al., 1999b) and an IRA for shrimp is ongoing.1 The risk of disease introduction with salmon for human consumption has also been considered by the New Zealand Government (MacDiarmid, 1994; Stone et al., 2001). These studies were undertaken for trade or regulatory purposes and thus have taken the commodity as the starting point, and the hazard identification considered all potential pathogens (since for a single commodity a number of pathogens may be identified). The report by Kahn et al. (1999a) demonstrates the depth and scale of data required for a comprehensive qualitative commodity import risk analysis. The starting point for two IRA was the hazard, G. salaris, a parasite of salmon. These assessments were not undertaken in response to a proposed commodity importation but to support the development of biosecurity and the direction of future research (Peeler and Thrush, 2004) and to assess the risk of disease spread following a change in EU legislation (Peeler et al., 2006). 1

http://www.daff.gov.au/corporate_docs/publications/pdf/biosecurityaustralia/bapm/2006/2006_35.pdf.

10

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

Table 4 Summary of disease spread aquatic animal risk analyses: risk analysis components completed Study

Risk analysis: components completed Hazard

Release

Exposure

Consequence

Risk management

Import risk analyses Kahn et al. (1999a) Kahn et al. (1999b) Peeler and Thrush (2004) Edgerton (2002) Bondad-Reantaso et al. (2005) Peeler et al. (2006) MacDiarmid (1994) Stone et al. (2001)

Yes Yes Yes Yes Yes Yes Yes Yes

Yes Yes Yes

Yes Yes Yes

Yes Yes Yes

Yes Yes

Yes Yes Yes Yes

Yes Yes

Yes

Yes

Yes

Yes

Other risk analyses Mortensen (2000) Bruneau (2001) LaPatra et al. (2001) Høga˚sen and Brun (2003) Paisley et al. (1999) Jansen et al. (2006) Peeler et al. (2004) Peeler et al. (2005) Munro et al. (2003)

Yes Yes Yes Yes Yes Yes Yes Yes Yes

Yes Yes Yes Yes Yes Yes Yes

Yes Yes Yes Yes Yes Yes Yes

Yes

The components of the risk analysis used varied considerably (Table 4). One study (Edgerton, 2002) did not progress beyond hazard identification and only five included a consequence assessment. In some cases the scope of the risk assessment will have been determined by the results obtained (if a negligible risk is found at the exposure or release stage, there is no justification for proceeding). The consequence assessment was, arguably, the weakest element of the IRA reviewed. The evidence on the susceptibility, including likely levels of mortality, of farmed and native wild fish species to the identified hazards was considered. However, the analyses did not include any quantitative assessments of the economic impact of disease to the aquaculture industry or the ecological impacts of declines of wild fish populations. 4.1.1. The prohibition of the introduction of imports of Canadian salmon into Australia One SPS dispute involving a fish commodity has reached the WTO dispute panel and appellate body; namely Canada’s challenge to Australia’s ban on imports of fresh, chilled or eviscerated salmon from the Northern hemisphere, which had been in place since 1975. The case illustrates the use of IRA in settling trade disputes and is worth some consideration. Australia claimed that a ban on salmon imports was necessary to protect its farmed salmon production from exotic diseases. The Australian Authorities had begun an IRA for salmon meat in the early 1990s (Beers and Wilson, 1993), possibly in anticipation of a formal challenge (Atik, 2004). In 1994, bilateral consultations on Canada’s request for access to the Australian market for its salmon began. The Australian Quarantine and Inspection Service (AQIS) commenced an IRA on non-viable salmon from North America,

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

11

which was produced in draft form in 1995 (recommending allowing the import of fresh salmon) and in final form in 1996 (recommending that the import ban remain) (DPIE, 1996). In 1997, the dispute was considered by a WTO dispute settlement panel and subsequently by the WTO Appellate body. The Appellate body of the WTO found that the risk assessment produced by Australia (DPIE, 1996) failed to appropriately evaluate either risk of introduction and establishment of exotic pathogens or their consequences. It concluded that the risk assessment only asserted that the possibility of disease introduction existed but did not, either qualitatively or quantitatively, evaluate the likelihood. Since a full evaluation of the likelihood of the hazards was lacking (Trachtman, 1999), the Australian measure was not based on a risk assessment. The Appellate body established three conditions to assess if a risk analysis was considered valid: (i) identification of the diseases that may be introduced and their consequences, (ii) evaluation of the likelihood of entry, establishment and spread of the diseases identified as hazards and their consequences and (iii) evaluation of the likelihood of entry, establishment and spread of the diseases according to the SPS measures that might be applied (WTO, 1998). Coherence between the risk assessment and the measures applied was emphasised. Secondly, it was found that Australia had not consistently applied its appropriate level of protection (in contravention of article 5.5 of the SPS agreement). The Appellate body found that other fish products (e.g. bait fish and ornamental fish) that carried the same or higher risks of disease introduction were imported with little or no restriction. The consequences of exotic disease introduction were potentially significant due to the size (and potential for expansion) of the salmon industry; however, it may also be argued that the existence of the salmon industry provides a reason for protectionist measures (Atik, 2004). The inconsistent treatment of different products appeared to reveal the protectionist motive underlying the ban on salmon imports. Following the WTO ruling AQIS produced a further IRA on non-viable salmonid and non-salmonid marine fish in 1999 (Kahn et al., 1999a). 4.2. Spread between wild and farmed fish populations The protection of wild populations is often a central objective of aquatic animal health management programmes. The use of risk analysis to underpin policies to protect wild fish is probably best illustrated by the case of G. salaris, a viviparous, monogenean, freshwater ecto-parasite of Atlantic salmon, whose natural hosts are Baltic strains of Atlantic salmon. It is regarded as a major threat to wild populations of Atlantic salmon, both in Norway and the UK. While the UK is free of G. salaris, the parasite was apparently introduced to Norway in early 1970s (with imports of juvenile salmon from Sweden) and spread to different geographical regions through stocking of rivers with infected fish (Johnsen and Jensen, 1991), with disastrous effects on the local stocks. Paisley et al. (1999) investigated the spread of G. salaris from sea cages situated in an estuary to the wild population in the river Tana. Other assessments have considered spread between river catchments in Norway (Høga˚sen and Brun, 2003; Jansen et al., 2006) and the UK (Peeler et al., 2004). Spread between wild and farmed populations was a crucial part of these studies, which were undertaken to support policies to minimise the risk of disease spread to wild populations. These studies have informed the development of policies to manage G. salaris in Norway

12

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

and the development of contingency plans and to set priorities for future research in the UK. 4.3. Biosecurity IRA has been used to support biosecurity at national or regional level by assessing the relative importance of different routes of introduction and spread, for example work on G. salaris was undertaken to support the development of both measures to minimise the likelihood of introduction (Peeler and Thrush, 2004) and spread (Peeler et al., 2004; Jansen et al., 2006). In principle, risk assessment could be used to assess routes of disease spread at any geographic scale (country, region, farm, or within farm) or to assess disease risks associated with a given practice, e.g. fish harvesting (Munro et al., 2003). The ranking of routes of entry and establishment of pathogens, either by qualitative or quantitative analysis, provides a sound basis for planning a biosecurity strategy (Peeler, 2005). At a farm level, risk analysis has been applied to investigate the transmission of vertically transmitted diseases between farms (via the sale of infected eggs) and within a farm (through an assessment of broodstock screening programmes) (Peeler et al., 2005).

5. Disease emergence An ‘‘emerging disease’’ is defined as a new disease, a new presentation of a known disease (e.g. increased severity, appearance in a new species) or an existing disease that appears in a new geographical area (Brown, 2000). Emerging diseases have become an increasingly important area of research in both human (Krause, 1998) and animal health (Daszak et al., 2001), including aquatic animals (Harvell et al., 1999). Murray and Peeler (2005) applied the CM model to explore the potential for disease emergence in aquaculture. A four stage risk model was developed: (i) emergence of a pathogen (release), (ii) establishment in a farmed population, (iii) establishment at the larger (regional) scale, and (iv) development of disease and its consequences (economic, ecological, welfare). The risk analysis approach provided a framework to break down the process of disease emergence and investigate its underlying processes (e.g. international movement of live animals, climate change, intensive fish production systems and interaction between wild and farmed fish). A qualitative risk assessment tool has been developed to assess the potential for disease emergence at an industry level (see Bridges et al. in this volume). Data on volume of trade and assessments of biosecurity (including contact with wild populations) as well as expert opinion were used. It was acknowledged that some important risk factors could not be included because they were not measurable.

6. Surveillance Risk analysis can improve the transparency and efficacy with which resources are allocated to different veterinary surveillance activities (Meah and Lewis, 1999), and the operational design of surveillance programmes (Sta¨rk et al., 2006). No current examples of

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

13

risk-based aquatic animal disease surveillance programmes were identified but the new EU Fish Health directive (2006/88/EC), proposes that statutory surveillance for notifiable fish disease be risk-based. It has been suggested that the term risk-based surveillance encompasses both the probability of the event occurring and the consequences (Sta¨rk et al., 2006), so the allocation of resources and the design of the programme is based on the potential losses due to disease. The application of risk analysis methods can be considered in a number of steps: (i) the selection of pathogens (the hazard identification stage), (ii) the selection of farms/rivers, (iii) selection of animals, and (iv) the sample size (Sta¨rk et al., 2006). It is likely that the surveillance programmes have in the past been designed by a similar process. However, risk analysis explicitly recognises the importance of consequence, as well as likelihood of disease occurrence, and provides a systematic, transparent approach to the design of surveillance programmes. The practical application of risk-based surveillance for aquatic animal diseases will require a thorough knowledge of the distribution of farmed and wild fish populations and the migration patterns of the latter. These data will need to be mapped within a geographic information system.

7. Further development of risk analysis Risk analysis reduces a complex biological process to a simple series of events or steps, it is therefore a form of modelling (Anderson and Nokes, 1997). Mathematical modelling of disease has been largely based on the hypothesis that the course of an epidemic depends on the rate of contact between susceptible and infectious individuals. In risk modelling, the release assessment examines the likelihood of the introduction of a hazard and establishment is examined in the exposure assessment. The likelihood of establishment of an introduced pathogen (i.e. whether an epidemic occurs) can be investigated through the application of mathematical modelling. Establishment will occur if the basic reproductive rate (R0) (defined as the average number of secondary cases of infection generated by one primary case in a susceptible population) is greater than one. Consequences will result if establishment occurs. The degree of consequences will largely be determined by the size of the epidemic (e.g. number of animals or farms affected and their geographic distribution). Qualitatively, different scenarios may be considered, e.g. rapidly identified and quickly eradicated, limited spread, etc. (Murray et al., 2004). Quantitative assessments require spatially explicit mathematical models to predict the extent of an outbreak and therefore the level of consequences. Mathematical disease modelling approaches, incorporated within a risk analysis framework, may improve estimates of the likelihood of exposure and severity of consequences. However, these modelling approaches require considerable amounts of data for accurate parameterisation. For example, to estimate R0 the number of susceptible individuals in the population, the duration of infectiousness of the primary case and the rate at which the primary case makes ‘effective’ contact (i.e. so that disease transmission occurs) with susceptibles are all required. A range of data will be required to estimate these parameters (e.g. population density, pathogen excretion, minimum infectious dose, contact rate). Estimating the likely extent of an outbreak requires spatially referenced data on population distributions and animal movements. However,

14

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

construction of these models will clearly identify gaps in the available data and thus priorities for future research. The consequences of disease introduction depend not only on the extent of spread of the pathogen but also its impact in terms of mortality and morbidity. Standard economic methods can be used to value direct and indirect economic effects of mortality and morbidity in farmed aquatic animals. Quantifying the environmental impact of the loss or decline of a wild aquatic animal population is more problematic. The term ‘non-use value’ (also known as passive use value) is used by economists to describe the value attributable to the simple knowledge that something, e.g. a wild fish population, exists. Methods such as contingent evaluation (Bennett and Larson, 1996) have been used to estimate non-use value. Wild fish may also have a ‘use-value’ if they are exploited (e.g. through recreational fishing). Non-use value was used in an economic study of wild stocks of the Snake River sockeye and chinook salmon (Olsen et al., 1991). Incorporating non-use value into consequence assessment goes some way to valuing, in economic terms, the impact of disease on biodiversity. Other mainstream epidemiological tools also have a role in risk analysis. Meta-analysis is a formal method for systematically finding and combining data sources (Dohoo et al., 2003), and therefore could be used in the evaluation of evidence in risk analysis.

8. Discussion In this paper we have attempted to review the role of risk analysis in aquatic animal health management, the challenges that exist and directions for future development. Risk analysis in fish health management has focused on disease spread to support management of international trade, disease transmission between wild and farmed populations, disease emergence and the development of surveillance. Stimulated by the SPS agreement, it has mainly been applied to pathogen hazards associated with international trade in live fish and fish products. Both qualitative and quantitative approaches have been used. Nearly all the IRA identified in this paper were undertaken in developed countries, despite the large majority of aquaculture production taking place in Asia, most notably China. It is possible that highly relevant and technically sound publications from Asia and Latin America have not been translated into English, and therefore not available from mainstream scientific literature. Organisations funding risk analyses should make efforts to make the work widely available (i.e. by translation and use of the Internet) to improve transparency and communication, two key characteristics of risk analysis. Risk analysis has also been applied more widely to support biosecurity (Peeler et al., 2004) and to support policies to minimise transmission between wild and farmed fish populations (Paisley et al., 1999; Høga˚sen and Brun, 2003). There are other single examples of the method being used to study disease emergence in aquaculture (Murray and Peeler, 2005) and vertical disease transmission (Peeler et al., 2005). In general the CM model, recommended by the O.I.E. for IRA, has been used and appears to be a highly flexible and appropriate approach to investigating diverse routes of disease spread. The choice between qualitative and quantitative approaches to risk analysis in aquatic animal health management will be determined by both the requirements of the decision

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

15

makers (i.e. is the qualitative result sufficient?), the availability of data and expertise of the analysts. IRA is generally used to justify sanitary measures. The SPS agreement stipulates that the assessed level of risk can only be reduced to, but not below, the acceptable level (WTO, 1995), which implies quantification of the risks (Pharo, 2003b), and thus the results of a qualitative IRA may not be appropriate for the task (although the O.I.E. guidelines for IRA state that both approaches are acceptable). However, a substantial range of data is required for a quantitative IRA (Pharo, 2003b) and it may be argued that in the absence of data, a quantitative analysis producing a result with very wide confidence limits is little, or no, better than a qualitative result. It is worth noting that the total uncertainty about the distribution of a parameter (e.g. duration of infectiousness) is due to true biological variability and our lack of knowledge (Vose, 2000). In aquatic systems, the problem is likely to be compounded by uncertainty about the level of variability (which complicates separation of these components). The additional benefit of a quantitative compared to a qualitative analysis should be judged by whether the basis for decision making is significantly improved. The evidence from the aquatic animal risk assessments reviewed indicated a relationship between the scale of the risk assessment and the availability of data. Two quantitative G. salaris risk analyses (Paisley et al., 1999; Høga˚sen and Brun, 2003) had a limited geographic scope, dealt with a single parasite and host. It is considerably more demanding to undertake a quantitative analysis for all pathogens associated with even a single commodity. In these cases a quantitative analysis of the most important hazard(s) may follow the initial qualitative analysis (MacDiarmid, 1994). Compared with terrestrial animals, a large number of ornamental fish species are traded internationally (Ariel, 2005) and information about diseases and parasites of these species is sparse (Mortensen, 2000; Hine, 2001). Thus, the hazard identification is problematic and unknown hazards are more likely to be frequently encountered for aquatic, compared with terrestrial animals (see Whittington and Chong in this volume). Article 5.7 of the SPS agreement allows members to take temporary, unsubstantiated SPS measures while seeking additional information ‘within a reasonable period of time’. It is probably unlikely that the fundamental information (e.g. for hazard identification) could be accumulated in a ‘reasonable period of time’. Factoring in a likelihood for unknown fish pathogen hazards, based on the frequency of the emergence of new diseases, has been suggested (Gaughan, 2002). It is unlikely that this option would be considered acceptable under the current SPS agreement. The SPS agreement requires that a risk assessment is an ‘‘evaluation of the likelihood of entry, establishment or spread of a pest or disease within the territory . . . and of the associated potential biological and economic consequences . . .’’. In the case of the Australian—measures affecting importation of salmon, the WTO appellate body found that the final report (DPIE, 1996) did meet this criteria. It may be argued that the available information did not allow for the degree of evaluation required by the Appellate body. The risk analyses reviewed focused on hazard identification, release and exposure assessment. In general, their consequence assessment components did not go beyond reviewing the evidence of species susceptibility for the identified hazards. The consequence assessment is a significant element of any risk analyses (risk being a function of both the likelihood and impact of a hazard) and is needed to evaluate the acceptable level of risk. However, compared with the release and exposure stages, the CM model (and OIE guidelines) provide little guidance on how consequence assessment should

16

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

be attempted. Consequence assessments, taking into account direct and indirect economic effects, require data on the spread of disease and the likely reaction of national and international markets (Pharo, 2003a). The loss or decline of a wild species may have a direct economic effect through the decline in recreational angling, i.e. user value, but ‘nonuser value’ would also have to be considered. A serious, quantitative assessment of consequences may well seem daunting for most risk analysts. The nature of consequence assessment highlights the need for multi-disciplinary approach (including economics and ecology). However, the data required for consequence assessment (e.g. the distribution of population, animal movements, contingent valuation of non-user value) may be frequently missing. For aquatic animals, spread of disease by wild populations needs to be considered, but data from disease outbreaks in wild populations will frequently not be available or be of limited use if they originate from a different environment. Disease modelling provides an approach to make best use of limited information and to identify parameters to which the outcome is most sensitive. The results can, therefore, be used to define further research. The use of consequence assessment within IRA might be made more transparent by guidelines which break consequences into a number of discrete steps: (i) likely size of the outbreak, (ii) the direct and indirect economic effects (including non-use value for wild populations), and (iii) the impact of disease at the ecosystem level. Every stage would not be required for all assessments. In addition to the introduction of exotic disease, government departments, responsible for natural resources, are concerned about the impact of aquaculture. Disease interaction between farmed and wild populations is, therefore, an important issue for aquatic animal health management. The G. salaris studies reviewed in this paper illustrate how risk analyses can support the development of government policy to protect wild fish stocks from the impact of fish farming. Decisions about managing G. salaris, raises a number of potentially controversial issues (e.g. allowing imports, expansion of production, stocking rivers with hatchery reared fish) and illustrates the need to balance the interests of aquaculture and the protection of wild populations. The perspectives (i.e. economic, social, environmental) of all stakeholders needs to be incorporated in the risk analysis (Stephen, 2001). The risk communication element of a risk analysis should ensure that stakeholders are consulted and informed. Risk analysis supports contentious decision making by the generation of an evidence base, which transparently documents uncertainty, information gaps and differentiates the types of information used (e.g. opinion, experimental results). Investigation of disease spread remains central to the application of risk assessment to the design of disease surveillance programmes. Risk analysis approaches can identify diseases and farms or animal strata within farms, which should be targeted on the basis of both probability (through analysis of routes of spread) and potential impact (i.e. consequence assessment) of infection. The demand for risk-based approaches to make best use of limited resources for surveillance is not currently matched by well developed methods (Sta¨rk et al., 2006). However, as in all applications of risk analysis, the evidence base of the analysis is crucial. An understanding of the routes of disease introduction and spread must underpin risk-based surveillance, and for aquatic animals transmission of pathogens via water currents and movements of escaped or wild fish must be a fundamental component. Therefore, an understanding of the ecology of susceptible wild fish populations (e.g. migration patterns, territories, etc.) and their density (and distribution

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

17

and the distribution of intermediate host species in the case of some parasites) must also be incorporated into risk-based surveillance for aquatic animal diseases. At a practical level this will involve integration of GIS with risk models. In many cases, information on disease transmission, especially in wild populations, may be lacking, and again, research needs to be oriented to generating the results to improve risk-based surveillance. Methodological advances are likely to take place in terrestrial animal disease risk-based surveillance. Policy makers responsible for aquatic animal health will need to adapt these methods to take account of the structure of the aquaculture industry, the spread of infection via water currents and the involvement of wild populations in transmission.

9. Conclusion Risk analysis confers a number of clear benefits in the assessment of scientific evidence in aquatic animal health decision making. Firstly, the requirement to document the evidence and assumptions used generates a degree of objectivity. Secondly, risk analysis incorporates risk communication which should allow all stakeholders to participate in the analysis, and thus build understanding around the conclusion and risk management decisions and gain a degree of ‘‘ownership’’ in the process. The quality of a risk analysis will ultimately depend on the skills of the risk analysis team and the available data. In this review we have highlighted a number of areas where data deficiencies limit the application of risk analysis. However, it can be argued that one important output of risk analyses is the identification of uncertain but important parameters, which can be used to direct research (to assess the true biological variability of a parameter). Epidemiologists have a large toolbox of methods at their disposal, we have attempted to identify how risk analysis can develop by the integration of other quantitative methods. In particular, the development of consequence assessment, in aquatic animal health, will require integration of disease modelling approaches and economic methods. The assessment of disease hazards associated with international trade in aquatic animals and their products is likely to remain the focus of risk analysis in the foreseeable future, although its application in other areas such as disease interaction between wild and farmed populations and surveillance is likely to increase. The level of protection to disease introduction offered by the SPS agreement, is likely to be a subject for continued debate (see Whittington and Chong in this volume). The case of the salmon imports into Australia clarified the WTO requirements for an IRA. The unresolved issue is how to achieve the required standard of IRA in the face of considerable data deficiencies. The development of more sophisticated methods is likely only to be a partial solution. In the longer term, more in depth research oriented towards the needs of risk analysis, is needed to improve the application of risk analysis in international trade and other areas.

Acknowledgements This work was undertaken as part of an EU framework 6 funded project: Disease interaction and pathogen exchange between wild and farmed aquatic animal populations—

18

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

a European network (DIPNET). We thank both our DIPNET colleagues and the reviewers for their constructive criticism. References Adams, S.M., Greeley, M.S., Ryon, M.G., 2000. Evaluating effects of contaminants on fish health at multiple levels of biological organization: extrapolating from lower to higher levels. Hum. Ecol. Risk Assess. 6, 15–27. Anderson, R., Nokes, D.J., 1997. Mathematical models of transmission and control. In: Detels, R., Holland, W.W., McEwen, J., Omenn, G.S. (Eds.), Oxford Textbook of Public Health. Oxford University Press, Oxford, pp. 689–719. Ariel, E., 2005. Ornamental fish as transboundary vectors of viral diseases. In: Walker, P.J., Lester, R.G., BondadReantaso, M.G. (Eds.), Diseases in Asian Aquaculture V. Fish Health Section, Asian Fisheries Society, Manila, Philippines, Queensland, Australia, pp. 103–112. Atik, J., 2004. The weakest link: demonstrating the inconsistency of ‘‘Appropriate Levels of Protection’’ in Australia-Salmon. Risk Anal. 24, 483–490. Beers, P.T., Wilson, D.W., 1993. Import risk assessment for salmon meat. Rev. Sci. Tech. Off. Int. Epiz. 12, 1147– 1152. Bennett, R.M., Larson, D.J., 1996. Contingent valuation of perceived benefits of farm animal welfare: an exploratory survey. J. Agric. Econ. 47, 224–235. Bondad-Reantaso, M., Lovell, E.R., Arthur, J.R., Hurwood, D., Mather, P.B., 2005. Pathogen and Ecological Risk Analysis for the Introduction of Blue Shrimp, Litopenaeus stylirostris, from Brunei Darussalam to Fiji. Secretariat of the Pacific Community. Brown, C., 2000. Emerging diseases of animals—an overview. In: Brown, C., Bolin, C. (Eds.), Emerging Diseases of Animals. American Society for Microbiology, p. 324. Bruneau, N., 2001. A quantitative risk assessment for the introduction of Myxobolus cerebralis to Alberta, Canada, through the importation of live salmonids. In: Rodgers, C.J. (Ed.), Risk Analysis in Aquatic Animal Health. Office des Epizooties Internationale (O.I.E.), Paris, pp. 41–50. Covello, V.T., Merkhofer, M.W., 1993. Risk Assessment Methods: Approaches for Assessing Health and Environmental Risks. Plenum Publishing, New York. Daszak, P., Cunningham, A.A., Hyatt, A.D., 2001. Emerging infectious diseases of wildlife—threats to biodiversity and human health. Science 287, 443–449. Dohoo, I.R., Martin, W., Stryhn, H., 2003. Veterinary Epidemiologic Research. Transcontinental Prince Edward Island, Canada, p. 706. DPIE, 1996. Salmon Import Risk Analysis: an assessment by the Australian Government of Quarantine controls on uncooked, Wild, Adult Ocean-Caught Pacific Salmonid Product sourced from the United States and Canada—Final report. Department of Primary Industries and Energy, Commonwealth of Australia, Canberra. Edgerton, B.F., 2002. Hazard analysis of exotic pathogens of potential threat to European freshwater crayfish. Bull. Fr. Peche Piscic. 367, 801–820. Gaughan, D.J., 2002. Disease-translocation across geographic boundaries must be recognized as a risk even in the absence of disease identification: the case with Australian Sardinops. Rev. Fish Biol. Fish. 11, 113–123. Harvell, C.D., Kim, K., Burkholder, J.M., Colwell, R.R., Epstein, P.R., Grimes, D.J., Hofman, E.E., Lipp, E.K., Osterhaus, A.D.M.E., Overstreet, R.M., Porter, J.W., Smith, G.W., Vasta, G.R., 1999. Emerging marine diseases—climate links and anthropogenic factors. Science 1505–1510. Hill, B.J., 1996. National Legislation in Great Britain for the control of fish diseases. Bulletin de l’Office International des Epizooties 15, 633–645. Hine, P.M., 2001. Problems of applying risk analysis to aquatic organisms. In: Rodger, C. (Ed.), Risk Analysis in Aquatic Animal Health. Office des Epizooties Internationale (O.I.E.), Paris, pp. 71–82. Høga˚sen, H.R., Brun, E., 2003. Risk of inter-river transmission of Gyrodactylus salaris by migrating Atlantic salmon smolts, estimated by Monte Carlo simulation. Dis. Aquat. Org. 57, 247–254. Jansen, P.A., Hogasen, H.R., Brun, E., 2006. Risk of spreading Gyrodactylus salaris in Norway, an evaluation of different potential pathways. In: Proceedings of the 11th Symposium of the International Society for Veterinary Epidemiology and Economics, Cairns, Australia, p. 213.

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

19

Johnsen, B.O., Jensen, A.J., 1991. The Gyrodactylus story in Norway. Aquaculture 98, 289–302. Kahn, S.A., Beers, P.T., Findlay, V.L., Peebles, I.R., Durham, P.J., Wilson, D.W., Gerrity, S.E., 1999a. Import Risk Analysis on Non-viable Salmonids and Non-salmonids Marine Finfish. Australian Quarantine and Inspection Service, Canberra, p. 409. Kahn, S.A., Wilson, D.W., Perera, R.P., Hayder, H., Gerrity, S.E., 1999b. Import Risk Analysis on Live Ornamental Finfish. Australian Quarantine and Inspection Service, Canberra, p. 172. Krause, R.M., 1998. Emerging Infections. Academic Press, New York. LaPatra, S.E., Batts, W.N., Overturf, K., Jones, G.R., Shewmaker, W.D., Winton, J.R., 2001. Negligible risk associated with the movement of processed rainbow trout, Oncorhynchus mykiss (Walbaum), from an infectious haematopoietic necrosis virus (IHNV) endemic area. J. Fish Dis. 24, 399–408. MacDiarmid, S.C., 1994. The Risk of Introducing Exotic Diseases of Fish into New Zealand Through the Importation of Ocean-caught Pacific Salmon from Canada. Ministry of Agriculture and Fisheries, New Zealand, Wellington, p. 161. Mackie, T.J., Arkwright, J.A., Pryce-Tannatt, T.E., Mottram, J.C., Johnstone, W.P., 1935. Final Report of the Furunculosis Committee. HMSO, Edinburgh, p. 67. Meah, M.N., Lewis, G.A., 1999. A Review of Veterinary Surveillance in England and Wales with Special Reference to Work Supported by the Ministry of Agriculture, Fisheries and Food—The Report to the Permanent Secretary. MAFF, London, p. 111. Mortensen, S., 2000. Scallop introductions and transfers, from an animal health point of view. Aquacult. Int. 8, 123–138. Munro, P.D., Murray, A.G., Fraser, D.I., Peeler, E.J., 2003. An evaluation of the relative risks of infectious salmon anaemia transmission associated with different salmon harvesting methods in Scotland. Ocean Coast. Manage. 46, 1–2. Murray, A.G., Peeler, E.J., 2005. A framework for understanding the potential for emerging diseases in aquaculture. Prev. Vet. Med. 67, 223–235. Murray, N., MacDiarmid, S.C., Wooldridge, M., Gummow, B., Morley, R.S., Weber, S.E., Giovannini, A., Wilson, D., 2004. Handbook on Import Risk Analysis for Animals and Animal Products—Introduction and Qualitative Risk Analysis. . O.I.E., 1993. Risk analysis and animal health and trade. Rev. Sci. Tech. Off. Int. Epiz. 12 (4), 448. O.I.E., 2001. Risk analysis in aquatic animal health. In: Rodgers, C.J. (Ed.), Proceedings of the OIE International Conference on Risk Analysis in Aquatic Animal Health, Paris. O.I.E., 2004. Aquatic Animal Health Code. Paris. Olsen, D., Richards, J., Scott, R.D., 1991. Existence and Sport Values for Doubling the Size of Columbia River Basin Salmon and Steelhead Runs. Rivers 2, 44–56. Paisley, L.G., Karlsen, E., Jarp, J., Mo, T.A., 1999. A Monte Carlo simulation model for assessing the risk of introduction of Gyrodactylus salaris to the Tana river, Norway. Dis. Aquat. Org. 37, 145–152. Peeler, E.J., 2005. The role of risk analysis and epidemiology in the development of biosecurity in aquaculture. In: Walker, P.J., Lester, R.G., Bondad-Reantaso, M.G. (Eds.), Diseases in Asian Aquaculture V. Fish Health Section, Asian Fisheries Society, Manila, Philippines, Queensland, Australia, pp. 35–46. Peeler, E.J., Gardiner, R., Thrush, M.A., 2004. Qualitative risk assessment of routes of transmission of the exotic fish parasite Gyrodactylus salaris between river catchments in England and Wales. Prev. Vet. Med. 64, 175–189. Peeler, E.J., Thrush, M.A., 2004. Qualitative analysis of the risk of introducing Gyrodactylus salaris into the United Kingdom. Dis. Aquat. Org. 103–113, 103–113. Peeler, E.J., Thrush, M.A., Midtlyng, P.J., Hill, B.H., 2005. Work Package 2 Report: The Application of Risk Assessment to the Study of Vertical Transmission of Fish Pathogens. VESO, Oslo, p. 28. Peeler, E., Thrush, M., Paisley, L., Rodgers, C., 2006. An assessment of the risk of spreading the fish parasite Gyrodactylus salaris to uninfected territories in the European Union with the movement of live Atlantic salmon (Salmo salar) from coastal waters. Aquaculture 258, 187. Pharo, H.J., 2003a. Economics in animal import risk assessment—can it deliver on its promise? In: Anon. (Eds.), Proceedings of the 10th International Symposium for Veterinary Epidemiology and Economics. Vina del Mar, Chile. Pharo, H.J., 2003b. The limits to quantification in import risk analysis. In: Anon. (Eds.), Proceedings of the 10th International Symposium for Veterinary Epidemiology and Economics. Vina del Mar, Chile.

20

E.J. Peeler et al. / Preventive Veterinary Medicine 81 (2007) 3–20

Pharo, H.J., MacDiarmid, S.C., 2001. Qualitative risk analysis of the risk of disease introduction through importation of salmon for human consumption. In: Rodgers, C.J. (Ed.), Risk Analysis in Aquatic Animal Health. Office des Epizooties Internationale (O.I.E.), Paris, pp. 51–60. Roberts, D., Orden, D., Josling, T., 1999. WTO disciplines on technical barriers to agricultural trade: progress, prospects and implications for developing countries. In: Proceedings of the Conference on Agricultural and the New Trade Agenda in the WTO Negotiations 2000 Geneva. Rodgers, C.J., 1997. Risk assessment in aquaculture report on a workshop held at the VIII EAFP international conference. In: Proceedings of the EAFP VIII International Conference. Herriot-Watt University, pp. 229– 233. Sta¨rk, K., Regula, G., Hernandez, J., Knopf, L., Fuchs, K., Morris, R., Davies, P., 2006. Concepts for risk-based surveillance in the field of veterinary medicine and veterinary public health: review of current approaches. Biomed. Central Health Service Res. 6 . Stephen, C., 2001. Role of risk assessment in fish health policy and management. ICES J. Mar. Sci. 58, 374–379. Stone, M.A.B., MacDiarmid, S.C., Pharo, H.J., 2001. Import health risk analysis. Salmonids for human consumption. In: Rodgers, C.J. (Ed.), Risk Analysis in Aquatic Animal Health. Office des Epizooties Internationale (O.I.E.), Paris, pp. 230–239. Trachtman, J.P., 1999. Decisions of the Appellate Body of the World Trade Organisation current survey. Australia—measures affecting importation of salmon. Eur. J. Int. Law 10, 192–195. Vose, D., 2000. Risk Analysis—A Quantitative Guide. John Wiley & Sons, Chichester. Vose, D., 2001. Qualitative versus quantitative risk analysis and modelling. In: Rodgers, C.J. (Ed.), Proceedings of the OIE International Conference on Risk Analysis in Aquatic Animal Health, Paris, pp. 19–26. WTO, 1995. Agreement on the Application of Sanitary and Phytosanitary Measures. World Trade Organisation, Geneva, p. 21. WTO, 1998. Australia – Measures Affecting Importation of Salmon – Report of the Appellate Body. WTO, Geneva. Zepeda, C., Salman, M., Ruppanner, R., 2001. International trade, animal health and veterinary epidemiology: challenges and opportunities. Prev. Vet. Med. 48, 261.