G Model
ARTICLE IN PRESS
PREVET-3549; No. of Pages 17
Preventive Veterinary Medicine xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed
Expert consultation on risk factors for introduction of infectious pathogens into fish farms Birgit C. Oidtmann a,∗ , Edmund J. Peeler a , Mark A. Thrush a , Angus R. Cameron b , R. Allan Reese a , Fiona M. Pearce a,c , Peter Dunn a , Trude M. Lyngstad d , Saraya Tavornpanich d , Edgar Brun d , Katharina D.C. Stärk e a
Centre for Environment, Fisheries and Aquaculture Science (Cefas), Barrack Road, Weymouth, Dorset DT4 8UB, United Kingdom AusVet Animal Health Services, 140 Falls Road, Wentworth Falls 2782, NSW, Australia c Ministry for Primary Industries, Pastoral House, 25 The Terrace, Wellington 6011, New Zealand d Norwegian Veterinary Institute, Pb 750 Sentrum, 0106 Oslo, Norway e Department of Production and Population Health, Royal Veterinary College, Hawkshead Lane, North Mymms, Hatfield AL9 7TA, United Kingdom b
a r t i c l e
i n f o
Article history: Received 9 August 2013 Received in revised form 18 March 2014 Accepted 20 March 2014
Keywords: Expert consultation Fish Pathogen introduction Risk factor Risk ranking
a b s t r a c t An expert consultation was conducted to provide quantitative parameters required to inform risk-based surveillance of aquaculture holdings for selected infectious hazards. The hazards were four fish diseases endemic in some or several European countries: infectious salmon anaemia (ISA), viral haemorrhagic septicaemia (VHS), infectious haematopoietic necrosis (IHN), and koi herpes virus disease (KHD). Experts were asked to provide estimates for the relative importance of 5 risk themes for the hazard to be introduced into and infect susceptible fish at the destination. The 5 risk themes were: (1) live fish and egg movements; (2) exposure via water; (3) on-site processing; (4) short distance mechanical transmission and (5) distance independent mechanical transmission. The experts also provided parameter estimates for hazard transmission pathways within the themes. The expert consultation was undertaken in a 2 step approach: an online survey followed by an expert consultation meeting. The expert opinion indicated that live fish movements and exposure via water were the major relevant risk themes. Experts were recruited from several European countries and thus covered a range of farming systems. Therefore, the outputs from the expert consultation have relevance for the European context. Crown Copyright © 2014 Published by Elsevier B.V. All rights reserved.
1. Introduction Animal health surveillance is conducted for several objectives, including the early detection of exotic, new and emerging diseases, demonstration of freedom from infection and monitoring disease prevalence (Doherr and Audigé, 2001; Stärk et al., 2006; Oidtmann et al., 2011b;
∗ Corresponding author. Tel.: +44 1305 206661; fax: +44 1305 206601. E-mail address: [email protected] (B.C. Oidtmann).
Cameron, 2012). Limited resources increase the need to improve the efficiency and effectiveness of surveillance activities. Risk-based surveillance (RBS) has the potential to increase the efficiency of resource allocation (Stärk et al., 2006). Whereas RBS approaches have been presented for a number of terrestrial animal diseases (trichinella, brucellosis, enzootic bovine leucosis, and avian influenza (Hadorn et al., 2002; Snow et al., 2007; Alban et al., 2008)), there are fewer examples for aquatic animal diseases. However, the application of these approaches to aquatic animal health
http://dx.doi.org/10.1016/j.prevetmed.2014.03.017 0167-5877/Crown Copyright © 2014 Published by Elsevier B.V. All rights reserved.
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
G Model PREVET-3549; No. of Pages 17
2
ARTICLE IN PRESS B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
is needed to improve the efficiency of surveillance (e.g. Gustafson et al., 2010; VHSV Expert Panel and Working Group, 2010; Oidtmann et al., 2011a, 2013). In Europe, aquaculture production businesses (APBs) producing fish to be marketed (from here on called fish farms or farms) are subject to European Council Directive 2006/88/EC on animal health requirements for aquaculture animals and their products, and on the prevention and control of certain diseases in aquatic animals (Anon., 2006, subsequently referred to as the ‘Directive’). Five fish diseases are currently notifiable under the Directive: infectious salmon anaemia (ISA), viral haemorrhagic septicaemia (VHS), infectious haematopoietic necrosis (IHN), koi herpes virus disease (KHD), and epizootic haematopoietic necrosis (EHN). The Directive requires regular farm inspections (to detect notifiable diseases, abnormal mortality and compliance with conditions of authorisation), the frequency of which should be determined by the disease status of the farm and the likelihood of pathogen introduction into and spread from the farm. Five disease categories (not to be confused with risk categories) for countries, zones or compartments are defined by the Directive: Category I – approved pathogen-free status; Category II – not declared disease-free, but subject to a surveillance programme to achieve disease-free status; Category III – infection status is unknown; Category IV – subject to an eradication programme, and Category V – where some farms (but not necessarily all) are known to be infected. Since there are multiple notifiable fish diseases, a single farm may be in multiple disease categories, depending on the pathogen (e.g. in Category I for VHS, and Category IV for IHN). The Directive requires that a risk-based approach is used for both disease surveillance (article 10 of the Directive) and compliance inspections (article 7). While a risk-based approach has been described in some countries for controls on food businesses (Maudoux et al., 2006; Lee et al., 2009; FAO, 2008), this approach is new in EU legislation for compliance controls of live animal holdings. The work presented here builds on earlier work for risk ranking of fish farms: Oidtmann et al. (2011a) developed a model for risk ranking fish farms that combined information on five main risk themes to obtain an integrated risk score for individual farms. The model development included expert consultation processes. However, the authors suggested using more rigorous elicitation methods to improve estimates to parameterise the model. Prior to the work presented here, we assessed the published literature, which showed that very little specific information was available to provide the quantitative parameters required by the risk model. We concluded that expert elicitation of parameters was required (Oidtmann et al., 2013) and conducted an expert consultation to support the parameterisation of a model for risk ranking fish farms in the EU. The parameter estimates were intended to be applicable across EU member states, to help set priorities for field visits by official inspectors for control purposes. The expert elicitation included the pathogens causing four fish diseases listed by the Directive as ‘nonexotic’ to Europe IHN virus (IHNV), VHS virus (VHSV), Koi Herpes Virus (KHV) and ISA virus (ISAV).
2. Material and methods The questionnaires were designed to provide parameter estimates that would be suitable to inform risk-based surveillance for selected pathogens based on risk of pathogen introduction to fish farms. In this paper the term risk is used as defined in epidemiology to indicate probability and not as used in risk analysis where it is defined to include both probability combined with consequences. A generic questionnaire was developed, covering the themes in Table 1, and adapted for the individual pathogens (see supplementary material) to take into account different exposure routes for marine or freshwater environments. Questionnaires for VHSV, IHNV, and KHV asked for estimates of pathogen transmission in the freshwater environment only and were identical except for the fish species the experts were asked to consider. The questionnaire for ISAV asked for estimates for pathogen transmission into farms in the marine environment. The first round of consultation was achieved by an online survey. The questionnaire was piloted with three experts whose first language was not English and minor modifications were made to make the questions clearer. Participants who were experts for more than one pathogen were invited to complete more than one questionnaire (e.g. most experts for VHS were also experts for IHN). Before completing the online survey, experts had been briefed on the context of the questionnaire to clarify the approach and ensure a common understanding of the questions. This was done by explaining in an email the purpose of the expert consultation and the concept of the hypothetical country (see below); a pdf copy of the questionnaire was provided. This was followed by a pre-arranged phone call to explain again the purpose of the expert consultation and the concept of the hypothetical country and to clarify any questions the experts may have had after going through the questionnaire. The link to the online questionnaire was sent to them following the phone call. The VHS questionnaire is shown in Appendix 1. Experts were asked to imagine a hypothetical country in Europe with 2000 farm sites,1 where the specified pathogen was present at a stated between-farm site level prevalence. Maps of the four scenarios were presented to the experts. The scenarios were: (1) 2% prevalence across the whole country; (2) 5% prevalence across the whole country; (3) and (4) assumed approved disease-free (i.e. Category I) zones within the country, but outside these zones farm sitelevel prevalence of 2 or 5%, respectively. The two different prevalence levels (2 or 5%) were used to explore whether prevalence influenced experts’ responses. The questionnaire was structured in 2 parts: part 1 aimed to allocate relative weights to the 5 risk themes for each scenario. This was achieved indirectly by asking the experts to imagine that 100 farm sites would become infected with the particular pathogen and to indicate how many of these sites had been infected via pathways in five risk themes over a 12-month period. They were asked to
1 In the questionnaire, the term farm site (rather than farm) was used (farms can consist of multiple farm sites).
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
G Model
ARTICLE IN PRESS
PREVET-3549; No. of Pages 17
B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
3
Table 1 Definitions of five themes for fish diseases used in an expert consultation on risk factors for introduction of infectious pathogens into fish farms. Theme name
Live fish and egg movements
Exposure via water
On-site processinga
Short distance mechanical transmission
Distance independent mechanical transmission
Theme definition
Introduction of pathogen through consignments of infected live fish (for ongrowing, stock enhancement or processing) and contaminated eggs from other sites, but not including pathogen present in transport water or contaminated transport tanks, lorries nets or packaging
Direct introduction of pathogen to farm source water by susceptible fish populations and activities upstream: i.e. other farmed and wild stocks (including fish released by farms for restocking or conservation purposes); fish processing
Introduction of pathogen through infected dead fish sourced from other sites for on-farm processing
Introduction of pathogen from sources in close proximity to the farm through routes including: piscivorous birds or other animals
Introduction of pathogen via contaminated fomites including: fish transporters and associated equipment; other equipment or personnel shared with other fish farms; fishery activities; visitors; vehicles, etc.
a
ISA experts were not consulted on theme ‘on-site processing’.
assume that all pathways were present on all farms. In scenarios 3 and 4 it was made clear that the farm sites were in approved disease-free zones. Part 2 of the questionnaire asked the experts to assume scenario 1 and allocate weights and probabilities for pathways within the 5 risk themes. Theme ‘processing on-site’ was omitted from the ISA questionnaire, since it was considered not relevant due to current management practices and regulations in Europe. Because of the context provided to the experts in our questionnaire, the VHSV genotype consulted on in our study was VHSV genotype Ia. The experts participating in the consultation are described in Tables 2a and 2b. Experts were mainly drawn from national fish health services for participating countries. Initial identification of experts was through personal contacts of the principal author (BO) and based on knowledge of experience in investigating outbreaks of notifiable fish diseases (field investigations or data analysis) or on their comprehensive research experience. Because EHN has not been reported in the EU, no expert consultation was undertaken for this disease. Questions were included in the introductory section of the questionnaire to assess the pathogen-specific expertise of the participants. One of the responses to the ISA online questionnaire was based on the agreed answers of 3 experts working in the same organisation. Four of the KHV experts were from the same organisation; their responses to the online questionnaire were combined into a single response (calculated mean).
2.1. Expert consultation meeting Experts who had participated in the online questionnaire were invited to a 2 day consultation meeting. Where multiple experts from a single organisation had contributed to the online questionnaire for a specific disease, responses had been combined into a single value, and only one expert per organisation and per disease was invited to the physical meeting to avoid bias. This meeting was structured as: (1) Group discussion I, (2) Plenary discussion I; (3) Group discussion II, and (4) Plenary discussion II.
Table 2a Expert consultation on risk factors for introduction of infectious pathogens into fish farms. Information about experts participating in the online questionnaire.
Total number of experts responding to online questionnaire Employment status Employed by non-government organisation Self-employed Employed by government organisation Describe your current work: • Analysis of fish farm production data • Collection of fish farm production data • Design of surveillance programmes for fish diseases • Implementation of fish health regulations/legislation • Analysis of data/information on the occurrence of disease in fish farms • Investigation of fish kills in wild populations • Laboratory diagnosis of fish disease • Aquatic animal health research • Collection of data/information on the occurrence of disease in fish farms • Advice to government on aquatic animal health • Investigation of disease outbreaks on fish farms
Number of experts
% of total
22
100
3
14
4 15
18 68
6
27
9
41
9
41
9
41
10
45
11
50
11
50
12
55
14
64
15
68
18
82
Frequent visits to fish farms? (more than 4 times per year) 9 No 13 Yes
41 59
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
G Model PREVET-3549; No. of Pages 17
ARTICLE IN PRESS B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
4
Table 2b Expert consultation on risk factors for introduction of infectious pathogens into fish farms. Responses to online questionnaire, country and experience of participating experts; expert attendance during workshop. Some experts were invited to respond to online questionnaires for more than 1 disease. VHVS Online questionnaire 15 • Experts invited to participate in online questionnaire (n) 12 • Responses (n) • Main place of work (country) of experts completing online questionnaires (n) Austria 2 Czech Republic 1 Denmark 1 England 1 Faroe Islands 0 Germany 2 Italy 1 Norway 0 Poland 1 Scotland 0 Switzerland 3 • Experience of experts responding to online questionnaire Years worked in aquatic animal health Mean 19.6 18 Median Min 2 Max 35 Number of investigations into outbreaks of the disease involved in over the past 10 years 25.6 Mean 26.5 Median 1 Min 60 Max Workshop • Experts attending workshop (n) • Experts in working group (n) a b
10 9
IHNV
KHV
ISAV
15 11
12 11 (8)a
8 5b
2 1 0 1 0 2 1 0 1 0 3
2 1 0 5 (2)a 0 2 0 0 1 0 0
0 0 0 0 2 0 0 4 (2)b 0 1 0
18.2 16 2 35
20.6 20 8 39
15.2 12 8 25
8.0 6 0 30
35.0 25 8 95
9.9 10 1 34
10 9
7 4
3 3
Responses from 4 experts combined to a single response (arithmetic mean). Three experts provided a single joint response and counted as 1 response.
Group discussion I: Experts were split into pathogenspecific break-out groups chaired by a facilitator. The concept behind the 4 scenarios in the questionnaires was introduced by the facilitator and clarification provided until the experts were confident that they understood the assumptions. The combined result of the responses from all experts for a given question from the online questionnaire was presented to the group in a boxplot. The experts were thus given the opportunity to consider the others’ opinions with the objective to agree on a consensus value from the whole expert group for each parameter (Delphi approach). The facilitator managed the discussion by: (1) identifying consensus when reached; (2) highlighting conflicting views; (3) finding compromise and (4) encouraging contributions from all group members. Plenary discussion I: The results from the group discussions were presented to all participants followed by discussion. Group discussion II: Previously compiled summaries of published information, relevant to assessing the scores for the various risk factors and themes for each pathogen were presented alongside the outputs from Group I discussions and experts were given the opportunity to revise scores. Plenary discussion II: Outcomes of the second group discussions were reported back to all participants. Experts from all groups therefore had an opportunity to comment, and query the results of other breakout groups. The detailed outcomes of the sessions and discussions were documented (available from the corresponding author on
request). The output included: agreed most likely, minimum and maximum values (single values agreed by the expert groups) for each of the parameters, with explanatory information where needed, such as identification of factors that may influence the parameter. 3. Results 3.1. Online questionnaire and expert consultation meeting On average, the experts had more than 15 years experience working in the field of aquatic animal health and, for those diseases widespread throughout Europe (VHS, IHN and KHV), most experts had investigated a substantial number of outbreaks of the disease for which they acted as experts (Tables 2a and 2b). All experts’ responses were used to calculate parameter estimates. 3.2. Parameter estimates obtained Results (median, minimum and maximum weights and likelihood estimates) from the expert elicitation are shown in Tables 3–5 (abbreviated questions; the full questions appear in Appendix 1). The weight medians do not always sum to 100 as they were calculated separately for each risk theme based on the responses provided by the experts. Some experts provided estimates in the initial online questionnaire that were clearly discordant with
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
Scenario 1
Risk theme
LFM
Water
Scenario 2
Pathogen VHSV Median Min Max
70 10 88
IHNV Median Min Max
Scenario 3
Scenario 4
Short
D-in
LFM
Water
Proc
Short
D-in
LFM
Water
Proc
Short
D-in
LFM
Water
Proc
Short
D-in
10 2 20
5 1 20
2.5 0 20
8 2 20
68 20 80
13 4 55
4 1 20
3.5 1 21
9 1 25
50 5 96
14 0 17
4 0 15
6 0 20
14 4 40
40 5 96
16 0 69
7.5 0 20
7.5 0 20
13 2 40
80 40 90
9 2 20
3 1 20
2 0 20
10 1 35
80 40 85
9 2 21
2 1 15
2 1 20
10 2 25
50 5 96
10 0 50
2 0 15
4 0 15
10 3 40
50 5 94
10 0 50
1 0 20
5 0 20
10 4 40
KHV Median Min Max
73 60 95
12 1 27
2 0 8
3 1 7
10 2 12
67 55 95
17 1 30
2 0 5
5 1 7
8 2 16
65 20 78
5 2 25
4 0 15
6 1 10
20 3 44
60 10 78
5 2 24
4 0 10
12 2 20
19 3 44
ISAV Median Min Max
5 1 9
50 40 60
2 1 3
42 15 40
8 3 13
45 35 55
2 1 3
45 40 50
60 10 99
20 10 25
20 0.3 40
60 10 99
20 10 25
0 0 0
0 0 0
20 0.3 40
Weights for risk themes assuming scenario 1–4. Scenario 1: 2% between farm site-level prevalence across the whole country; scenario 2: 5% between farm site-level prevalence across the whole country; scenarios 3 and 4: assumed approved disease-free zones within the country, but outside these zones between farm site-level prevalence of 2 or 5%, respectively. Pathogens: VHSV, viral haemorrhagic septicaemia virus; IHNV, infectious haematopoietic necrosis virus; KHV, koi herpes virus; ISAV, infectious salmon anaemia virus. Risk themes: LFM, live fish and egg movements; Water, exposure via water; Proc, on-site processing; Short, short distance mechanical transmission; D-in, distance independent mechanical transmission.
Table 4 Expert consultation on risk factors for introduction of infectious pathogens into fish farms. Estimates of likelihood of transmission of pathogen with live fish movements from a subclinically infected source site to a receiving rainbow trout (VHS, IHN), carp (KHV) or Atlantic salmon (ISA) farm site, by species. IHNV
VHSV
KHV
ISAV
Species
Median
Min
Max
Species
Median
Min
Max
Species
Median
Min
Max
Species
Median
Min
Max
Rainbow trout Brown Trout Brook Trout Grayling Pike
80 30 7.5 37.5 0.5
10 1 0 0 0
100 85 10 60 1
Rainbow trout Atlantic salmon Brown Trout Brook Trout Pike
80 60 5 5 5
10 2 0 0 0
95 95 10 10 10
Common Carp Goldfish Grass Carp Tench
70 20 20 10
30 2 0.75 0.75
90 40 20 20
Atlantic salmon Rainbow trout
95 10
90 5
100 15
ARTICLE IN PRESS
Proc
G Model
PREVET-3549; No. of Pages 17
Scenario
B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
Table 3 Expert consultation on risk factors for introduction of infectious pathogens into fish farms.
5
G Model PREVET-3549; No. of Pages 17
6
ARTICLE IN PRESS B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
the majority. Reasons for such deviations were clarified in all cases by discussion (misinterpretation of the question or regional differences). Following clarification, the group estimates were agreed by all experts in the specific working groups and approved by the plenary. No changes were made by any of the groups based on reflection of the published information (Group discussion II). Median estimates are used in the sections below unless stated otherwise.
3.2.1. Weights for themes (Table 3) For VHSV, IHNV, and KHV, live fish movements was the most important risk theme for all scenarios (Table 3). Live fish movements were considered less relevant for farm sites located in zones with Category I status (approved pathogen-free) in scenarios 3 and 4, compared to farm sites located in an endemically infected country, although it remained the most important risk theme. Experts assumed compliance with current regulations, i.e. farmers in Category I zones would source fish from farms of equivalent health status. With higher assumed prevalence (scenario 2 and 4 vs. scenario 1 and 3 respectively), the scores for live fish movements decreased, while they increased for transmission via water or short distance mechanical transmission (VHSV and KHV). The range of values (minimum to maximum) for some themes and diseases was both wide and skewed (e.g. live fish movement for scenario 1 VHS has a median of 80, a minimum value of 10, and a maximum of 88). For the seawater disease pathogen (ISAV), exposure via water and distance-independent mechanical transmission were the most important themes for scenarios 1 and 2. Live fish movements were rated the most important risk theme for scenarios 3 and 4. Farm sites located in Category I zones were considered to be at a lower risk via mechanical transmission and water (compared to farms located in an endemically infected country – scenarios 1 and 2), which increased the importance of other themes, notably live fish movements. The minimum to maximum ranges were lower than for VHSV, IHNV and KHV.
3.2.2. Likelihood of transmission of pathogen with live fish movements (Table 4) The experts were asked to estimate the likelihood of transmitting pathogen through live fish movements from a subclinically infected source site by different species. The likelihood estimates (medians) provided for a single movement ranged from 70% (carp, KHV) to 95% (Atlantic salmon, ISAV) (Table 4). The ranges (min to max) around these median values were very wide for VHSV, IHNV and KHV (e.g. 10–100 for VHSV transmission by rainbow trout) but narrow for ISAV transmission by Atlantic salmon (90–100). For VHSV, IHNV and KHV, experts were asked to consider species not currently listed as susceptible to these diseases by the Directive (i.e. brook trout for VHSV; brown trout, brook trout and pike for IHNV; goldfish, grass carp, and tench for KHV). The medians for these species ranged from 5 to 20.
3.2.3. Scores for pathways within the risk themes (Table 5) The results presented in Table 5 cover questions 26–44 of the VHSV questionnaire (detailed estimates for specific pathways within the 5 risk themes) and the corresponding questions within the questionnaires for the other pathogens. The experts were asked to imagine a hypothetical country with 2000 farm sites and 2% prevalence between farm sites for each specific pathogen. Under the assumption that the source site was subclinically infected, the experts estimated that of 100 consignments of non-disinfected rainbow trout eggs, 40 would lead to infections of VHSV and 30 of IHNV at receiving sites. Of 100 consignments of Atlantic salmon eggs, 1 would lead to infection of ISAV at receiving sites. If eggs were disinfected, the experts reduced these estimates to 1 (VHSV), 5 (IHNV) or 0.01 (ISAV) (Q26). This question was omitted for KHV, since transport of fish eggs between sites is not common practice for cyprinid aquaculture. Within theme ‘exposure via water’, presence of farms holding species susceptible to infection with the specific pathogen within 5 km upstream/sea-distance was considered the highest risk (Q30). Estimates were that over a 1 year period, out of 100 independent farm sites (located in independent water catchments/sea water zones), there would be 32.5 infections of VHSV, 20 of IHNV, 30 of KHV or 20 of ISAV if an infected farm was located within 5 km upstream/seaway distance (Q27). The number of farm sites estimated to become infected over a 1 year period as a result of the presence of infected wild fish within 5 km of these farm sites was 10 (VHSV and IHNV), 15 (KHV) and 5 (ISAV) (Q28). For VHSV and IHNV, the risk of infection from the presence of a fish processing facility within 5 km upstream was rated similar to the presence of wild fish populations or stocked fisheries (susceptible species) within 5 km upstream (Q30). On-site processing was considered an important risk for VHSV, IHNV, and KHV introduction (Q32–36). The experts estimated that out of 100 farm sites receiving rainbow trout (or carp) carcases for processing from other farm sites (of unknown infection status), 5 would become infected over a 12-month period (VHSV, IHNV, KHV) (Q33). When the processing site received live infected fish, the estimates increased to 77.5 (VHSV), 77 (IHNV) and 80 (KHV) (Q36). The risk of infection due to short distance mechanical transmission (assuming no direct water connectivity) was estimated to be less than 6 of 100 farms per year across all scenarios and diseases (Q37). The highest estimates were for KHV, 4.6 (scenario I), 5.8 (scenario 2), 2.5 (scenario 3) and 3.8 (scenario 4) of 100 farms, compared to 3 or fewer sites for VHSV, and 2 or fewer for IHNV. The higher the prevalence in the country, the higher the number of farm sites expected to become infected. The experts for ISAV estimated that no farm sites in a declared disease-free zone would become infected due to short distance mechanical transmission, and 0.5 or less of 100 farm sites to become infected in an endemically infected country outside a disease-free zone. In ranking three potential sources for infection via this route (Q38), fish farms tended to be considered the most relevant, then fisheries
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
Q
Topic of question
VHSV Median
IHNV Min
Max
Median
KHV Min
Max
Median
ISAV Min
Max
26
Eggs not disinfected Eggs disinfected
40 1
0 0
100 5
30 5
1 0
98 10
N/A N/A
Exposure via water
27 28 29
Farm Wild fish Fisherya
5 0.5 1
100 50 100
80 20 80
80 50 60
Farm Processor
30 15 30 N/A 50 5
1 2 15
Transport route (ISA) Weight by source
20 10 12 N/A 50 20
1 0.5 1
30
32.5 10 20 N/A 50 25
20 0
80 20
Wild/fisheryb Transport route (ISA) Other
5
60
0
20 55 100 10 70 50 95
On-site processing
Short distance mechanical transmission
Distance independent mechanical transmission
a
c d e f
17.5 N/A 5
5
45
0
10
40 47.5 5 27.5 10 77.5
1 3 0 5 1 10
80 93 50 100 50 100
Scenario 1 Scenario 2 Scenario 3 Scenario 4 Farms Fisheries Wild 5 hae
33 39
Within 2 km
Min 0 0
Max 2 0.02
20 5 10 5 79 5
5 1 2 2 20 2
40 10 20 10 100 10
10 5 1
2 2 0
20 10 2
40 5
75 42
5
45
0
10
36 N/A 9
30 49 5 20 10 77
1 3 1 5 1 15
90 99 10 90 50 99
40 60 5 15 20 80
0 45 0 3 1 10
5 20 5 3 75 40 40 60
1 2 0.5 1 60 30 10 33
0 0 0 0 20 20 1 5
5 4 2 3 70 40 40 60
4.6 5.8 2.5 3.8 40 40 20 28.3
0.1 0.2 0.1 0.1 20 20 2 10
13.7 20.5 27.4 38.4 85 75 40 60
0.1 0.5 0 0 95 N/A 5 N/A
0 0 0 0 90 2
10
20 10
45 75
33 33
20 10
45 75
35 35.9
30 10
45 60
40 60
30 50
50 75
60 25 15
40 5 1
94 35 30
65 25 10
40 5 1
94 35 30
60 27.5 10
20 20 2
85 45.8 40
75 20 5
60 10 2
80 40 10
19 N/A 1
Process No processing Dead unknown Dead infected Live unknown Live infected
37
No. farms infected
38
Weight by source
39
Proportion farms inf.
40
Fishery/farmf
41
Scenario 1 Scenario 2 Scenario 3 Scenario 4 Staff Equipment Fish vehicles General vehicles Fishery Unauth people Unauth vehicles Insufficient dis vehicles Insufficient dis footwear Staff
2.5 3.5 1 2 20 35 22.5 5 10 7.5 5 30 30 30
1 1 0 0 5 5 8 0 2 0 0 10 10 5
10 10 3 5 30 50 50 20 30 15 10 40 40 70
2 4 1 2.5 20 35 25 5 10 5 5 30 30 30
1 1 0 0.5 5 5 8 0 2 0 0 10 10 5
10 30 5 5 30 40 50 10 25 15 15 40 40 70
7.3 13.8 4.7 5.4 12.5 37.5 20 1 12.5 5.6 6.9 30 30 30
1 1 1 2 3 15 5 0 5 1 0 0 5 5
21 30 29 29 30 45 40 10 75 15 10 60 35 93
2 10 1 5 2 30 60 8 N/A 1 1 80 1 17
1 2 0 2 1 20 50 2
4 20 2 10 4 45 75 20
0 0 75 0 10
2 2 95 2 25
Avian Mammalian Stored waste Staff
20 7.5 40 22.5
1 1 20 1
50 30 80 65
20 5 40 26
1 1 20 5
50 30 80 65
20 10 40 24.9
2 2 20 5
45 30 70 60
4 1 80 15
1 0 70 5
8 2 95 30
43
b
75 45
1 0.01
33 34 35 36
32
42
Short distance and distance independent mechanical transmission
30 5
Median
44
0.2 1 0 0 100
For ISAV the experts provided an estimate for presence of processing site within 5 km. For ISAV: the experts provided an estimate for presence of wild fish only within 5 km. For ISAV: proportion of farms with 1,000,000 fish on site. For ISAV: the experts provided an estimate for presence of farm sites within 5 km.
ARTICLE IN PRESS
Live fish and egg movements
G Model
PREVET-3549; No. of Pages 17
Theme
B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
Table 5 Expert consultation on risk factors for introduction of infectious pathogens into fish farms. Relative weights or likelihood estimates for pathogen transmission for individual risk pathways within risk themes. The number in column Q refers to the question no. in the VHS questionnaire. A keyword indicates the subject of the question.
7
G Model
ARTICLE IN PRESS
PREVET-3549; No. of Pages 17
B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
8
Table 6 Expert consultation on risk factors for introduction of infectious pathogens into fish farms. Q
VHSV
IHNV
KHV
ISAV 1 15 30 4
42
Staff Equipment Fish vehicles General vehicles Fishery
1 1.75 1.13 0.25 0.5
1 1.75 1.25 0.25 0.5
1 3 1.6 0.08 1
43
Unauthorised people Unauthorised vehicles Insufficiently disinfected vehicles Insufficient disinfected footwear Staff
0.25 0.17 1
0.17 0.17 1
0.19 0.23 1
0.06 0.06 4.71
1
1
1
0.06
1
1
1
1
Avian Mammalian Stored waste Staff
0.89 0.33 1.78 1
0.77 0.19 1.54 1
0.80 0.40 1.61 1
0.27 0.07 5.33 1
44
Relative importance of mechanical transmission pathways of VHSV, IHNV, KHV and ISAV. Medians of estimates provided by experts to questions 42–44 on individual mechanical transmission pathways scaled against transmission via staff. The number in column Q refers to the question no. in the VHS questionnaire. A keyword indicates the mechanical transmission pathway.
and thirdly wild fish populations. Farm size was considered to have relatively little impact on the likelihood of a farm becoming infected through short distance mechanical transmission (Q39). Distance to other farms or fisheries was highly relevant (Q40). For distance independent mechanical transmission the estimates were consistently highest for KHV across all scenarios and highest for scenario 2 across all diseases (Q41). Mechanical transmission routes were compared in three separate questions (Q42–44). Risk of transmission via staff was included in all three so that responses could be scaled against this pathway for each question. Shared equipment and storing fish waste on site were identified as the highest risks for mechanical transmission of VHSV and IHNV. Vehicles were considered the most relevant route for ISAV, and equipment most relevant for KHV (Table 6). Details of adjusted most likely values and discounted outliers emerging from the expert consultation meeting compared to the online questionnaire results are available from the corresponding author on request. 4. Discussion The use of expert opinion is an accepted approach where information is not available (Gustafson et al., 2013). Expert elicitation has been used in the context of diseases in terrestrial animals and also for fish diseases (Bruneau et al., 1999; Gustafson et al., 2005, 2010, 2013). We used a twostep approach based on Delphi technique (Dalkey, 1969). A facilitator managed discussions with defined instructions to ensure participation from all experts, previously recognised as a potential issue (Oidtmann et al., 2011a). Experts were given four scenarios to explore the importance of disease prevalence and status. An initial setting in which the country was already infected was chosen to avoid a focus on pathways of introduction into areas
previously free of the disease. A prevalence of 2% was chosen, because it is a commonly used design prevalence for the first level of clustering (e.g. proportion of infected farm sites in a zone) for animal diseases when demonstrating freedom from disease (Corsin et al., 2009; World Organisation for Animal Health OIE, 2010). Thus scenario 1 best reflects the potential situation in a country that wishes to demonstrate freedom from infection. Scenarios 3 and 4 were used to identify whether there were differences in importance of the routes depending on whether a farm site was located in a country aiming to achieve disease-free status opposed to a farm site in a zone. The format of the questionnaire was novel to the majority of experts and some found it difficult to understand the questions asked. However, following the expert group work, there was a consensus amongst all experts that the values agreed during the group sessions were appropriate. In all groups, consensus was achieved regarding the most likely values for all parameters. Consensus was assessed by inviting experts individually to comment on the most likely values and the range (minimum and maximum values) for risk pathways in the Group discussions. Particular attention was paid to obtaining comments from experts who were less active in group discussions. Variability in the responses to the online questionnaire (minimum and maximum values, Tables 3–5) usually reflected the different context and industry structure in the countries represented. Experts agreed during the meeting that the values provided in the online questionnaire were suited to describe the range these values could take (small modifications were made during Group discussions, where an outlier stemmed from a misunderstanding of a question). For the remote part of the expert consultation (the online questionnaire), individual experts had not been asked to quantify a range around an estimate. Ranges were obtained by summarising the responses from all experts for a specific pathogen to a given question. Given the paucity of information and the widely differing production systems for the same species across the member states represented, it is not surprising that the estimates of the weights and likelihood estimates had wide ranges. During discussions in the meeting it emerged that some of the variation in experts’ scores is explained by the variation in level of exposure to risk pathways between regions. However, the experts agreed that the medians were the appropriate most likely values and that the lowest and highest estimates provided by experts in the online questionnaire suitably represented the appropriate range the respective estimates could take. Several of the experts’ estimates, such as likelihood of transmission of virus via subclinically infected fish, are independent of regional practices and therefore have relevance across the board for the given pathogen. Risk of pathogen introduction to a farm site is dependent on whether a certain exposure pathway applies and its frequency. Therefore, identification of farms at higher risk of becoming infected with a specific pathogen is ideally informed by the presence or absence of certain risk pathways and the level of exposure to this pathway if it applies. Formulas can be devised to take level of exposure into account (e.g. number of live fish introductions to a given farm site; number of movements of vehicles or equipment
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
G Model PREVET-3549; No. of Pages 17
ARTICLE IN PRESS B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
– with/without disinfection – onto a site, etc.). However, there may also be scope for further narrowing the ranges appropriate for a certain region through further expert consultation using local experts. The approach used in this expert consultation followed several recommendations recently summarised by Gustafson et al. (2013): experts were individually contacted and the questionnaire discussed with them before they were asked to respond, which allowed minor adjustments to the questionnaire or further explanations. Leading questions were avoided (with the intention of avoiding anchoring) and counts relating to clearly specified denominators were elicited rather than probabilities. Experts then met for group discussions to share the results of the online questionnaire and discuss the distribution of the responses. Weighting of responses has been suggested to reflect different levels of experience of participants in an expert consultation (Cooke, 1991; Cooke and Goossens, 2000; Aspinall, 2010). However, peer and self-evaluation may be unreliable predictors of expert performance. More accurate estimates from expert consultations can result from systems that provide feedback about experts’ performance, encourage experts to think about their estimates and aggregate individual estimates (Cooke, 1991; Fisher, 2009; Burgman et al., 2011). In fact, expert advice will be more accurate if based on broadly defined expert groups, structured question protocols and feedback (Burgman et al., 2011). These principles we have tried to apply in this study. 4.1. Relevance of live fish movements and exposure via water
9
diseases (e.g. brook trout (Salvelinus fontinalis) for both VHS and IHN) were nevertheless considered to carry a significant risk for pathogen transmission. 4.3. On-site processing On-site processing received a relatively low weight (
ARTICLE IN PRESS
PREVET-3549; No. of Pages 17
Preventive Veterinary Medicine xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed
Expert consultation on risk factors for introduction of infectious pathogens into fish farms Birgit C. Oidtmann a,∗ , Edmund J. Peeler a , Mark A. Thrush a , Angus R. Cameron b , R. Allan Reese a , Fiona M. Pearce a,c , Peter Dunn a , Trude M. Lyngstad d , Saraya Tavornpanich d , Edgar Brun d , Katharina D.C. Stärk e a
Centre for Environment, Fisheries and Aquaculture Science (Cefas), Barrack Road, Weymouth, Dorset DT4 8UB, United Kingdom AusVet Animal Health Services, 140 Falls Road, Wentworth Falls 2782, NSW, Australia c Ministry for Primary Industries, Pastoral House, 25 The Terrace, Wellington 6011, New Zealand d Norwegian Veterinary Institute, Pb 750 Sentrum, 0106 Oslo, Norway e Department of Production and Population Health, Royal Veterinary College, Hawkshead Lane, North Mymms, Hatfield AL9 7TA, United Kingdom b
a r t i c l e
i n f o
Article history: Received 9 August 2013 Received in revised form 18 March 2014 Accepted 20 March 2014
Keywords: Expert consultation Fish Pathogen introduction Risk factor Risk ranking
a b s t r a c t An expert consultation was conducted to provide quantitative parameters required to inform risk-based surveillance of aquaculture holdings for selected infectious hazards. The hazards were four fish diseases endemic in some or several European countries: infectious salmon anaemia (ISA), viral haemorrhagic septicaemia (VHS), infectious haematopoietic necrosis (IHN), and koi herpes virus disease (KHD). Experts were asked to provide estimates for the relative importance of 5 risk themes for the hazard to be introduced into and infect susceptible fish at the destination. The 5 risk themes were: (1) live fish and egg movements; (2) exposure via water; (3) on-site processing; (4) short distance mechanical transmission and (5) distance independent mechanical transmission. The experts also provided parameter estimates for hazard transmission pathways within the themes. The expert consultation was undertaken in a 2 step approach: an online survey followed by an expert consultation meeting. The expert opinion indicated that live fish movements and exposure via water were the major relevant risk themes. Experts were recruited from several European countries and thus covered a range of farming systems. Therefore, the outputs from the expert consultation have relevance for the European context. Crown Copyright © 2014 Published by Elsevier B.V. All rights reserved.
1. Introduction Animal health surveillance is conducted for several objectives, including the early detection of exotic, new and emerging diseases, demonstration of freedom from infection and monitoring disease prevalence (Doherr and Audigé, 2001; Stärk et al., 2006; Oidtmann et al., 2011b;
∗ Corresponding author. Tel.: +44 1305 206661; fax: +44 1305 206601. E-mail address: [email protected] (B.C. Oidtmann).
Cameron, 2012). Limited resources increase the need to improve the efficiency and effectiveness of surveillance activities. Risk-based surveillance (RBS) has the potential to increase the efficiency of resource allocation (Stärk et al., 2006). Whereas RBS approaches have been presented for a number of terrestrial animal diseases (trichinella, brucellosis, enzootic bovine leucosis, and avian influenza (Hadorn et al., 2002; Snow et al., 2007; Alban et al., 2008)), there are fewer examples for aquatic animal diseases. However, the application of these approaches to aquatic animal health
http://dx.doi.org/10.1016/j.prevetmed.2014.03.017 0167-5877/Crown Copyright © 2014 Published by Elsevier B.V. All rights reserved.
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
G Model PREVET-3549; No. of Pages 17
2
ARTICLE IN PRESS B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
is needed to improve the efficiency of surveillance (e.g. Gustafson et al., 2010; VHSV Expert Panel and Working Group, 2010; Oidtmann et al., 2011a, 2013). In Europe, aquaculture production businesses (APBs) producing fish to be marketed (from here on called fish farms or farms) are subject to European Council Directive 2006/88/EC on animal health requirements for aquaculture animals and their products, and on the prevention and control of certain diseases in aquatic animals (Anon., 2006, subsequently referred to as the ‘Directive’). Five fish diseases are currently notifiable under the Directive: infectious salmon anaemia (ISA), viral haemorrhagic septicaemia (VHS), infectious haematopoietic necrosis (IHN), koi herpes virus disease (KHD), and epizootic haematopoietic necrosis (EHN). The Directive requires regular farm inspections (to detect notifiable diseases, abnormal mortality and compliance with conditions of authorisation), the frequency of which should be determined by the disease status of the farm and the likelihood of pathogen introduction into and spread from the farm. Five disease categories (not to be confused with risk categories) for countries, zones or compartments are defined by the Directive: Category I – approved pathogen-free status; Category II – not declared disease-free, but subject to a surveillance programme to achieve disease-free status; Category III – infection status is unknown; Category IV – subject to an eradication programme, and Category V – where some farms (but not necessarily all) are known to be infected. Since there are multiple notifiable fish diseases, a single farm may be in multiple disease categories, depending on the pathogen (e.g. in Category I for VHS, and Category IV for IHN). The Directive requires that a risk-based approach is used for both disease surveillance (article 10 of the Directive) and compliance inspections (article 7). While a risk-based approach has been described in some countries for controls on food businesses (Maudoux et al., 2006; Lee et al., 2009; FAO, 2008), this approach is new in EU legislation for compliance controls of live animal holdings. The work presented here builds on earlier work for risk ranking of fish farms: Oidtmann et al. (2011a) developed a model for risk ranking fish farms that combined information on five main risk themes to obtain an integrated risk score for individual farms. The model development included expert consultation processes. However, the authors suggested using more rigorous elicitation methods to improve estimates to parameterise the model. Prior to the work presented here, we assessed the published literature, which showed that very little specific information was available to provide the quantitative parameters required by the risk model. We concluded that expert elicitation of parameters was required (Oidtmann et al., 2013) and conducted an expert consultation to support the parameterisation of a model for risk ranking fish farms in the EU. The parameter estimates were intended to be applicable across EU member states, to help set priorities for field visits by official inspectors for control purposes. The expert elicitation included the pathogens causing four fish diseases listed by the Directive as ‘nonexotic’ to Europe IHN virus (IHNV), VHS virus (VHSV), Koi Herpes Virus (KHV) and ISA virus (ISAV).
2. Material and methods The questionnaires were designed to provide parameter estimates that would be suitable to inform risk-based surveillance for selected pathogens based on risk of pathogen introduction to fish farms. In this paper the term risk is used as defined in epidemiology to indicate probability and not as used in risk analysis where it is defined to include both probability combined with consequences. A generic questionnaire was developed, covering the themes in Table 1, and adapted for the individual pathogens (see supplementary material) to take into account different exposure routes for marine or freshwater environments. Questionnaires for VHSV, IHNV, and KHV asked for estimates of pathogen transmission in the freshwater environment only and were identical except for the fish species the experts were asked to consider. The questionnaire for ISAV asked for estimates for pathogen transmission into farms in the marine environment. The first round of consultation was achieved by an online survey. The questionnaire was piloted with three experts whose first language was not English and minor modifications were made to make the questions clearer. Participants who were experts for more than one pathogen were invited to complete more than one questionnaire (e.g. most experts for VHS were also experts for IHN). Before completing the online survey, experts had been briefed on the context of the questionnaire to clarify the approach and ensure a common understanding of the questions. This was done by explaining in an email the purpose of the expert consultation and the concept of the hypothetical country (see below); a pdf copy of the questionnaire was provided. This was followed by a pre-arranged phone call to explain again the purpose of the expert consultation and the concept of the hypothetical country and to clarify any questions the experts may have had after going through the questionnaire. The link to the online questionnaire was sent to them following the phone call. The VHS questionnaire is shown in Appendix 1. Experts were asked to imagine a hypothetical country in Europe with 2000 farm sites,1 where the specified pathogen was present at a stated between-farm site level prevalence. Maps of the four scenarios were presented to the experts. The scenarios were: (1) 2% prevalence across the whole country; (2) 5% prevalence across the whole country; (3) and (4) assumed approved disease-free (i.e. Category I) zones within the country, but outside these zones farm sitelevel prevalence of 2 or 5%, respectively. The two different prevalence levels (2 or 5%) were used to explore whether prevalence influenced experts’ responses. The questionnaire was structured in 2 parts: part 1 aimed to allocate relative weights to the 5 risk themes for each scenario. This was achieved indirectly by asking the experts to imagine that 100 farm sites would become infected with the particular pathogen and to indicate how many of these sites had been infected via pathways in five risk themes over a 12-month period. They were asked to
1 In the questionnaire, the term farm site (rather than farm) was used (farms can consist of multiple farm sites).
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
G Model
ARTICLE IN PRESS
PREVET-3549; No. of Pages 17
B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
3
Table 1 Definitions of five themes for fish diseases used in an expert consultation on risk factors for introduction of infectious pathogens into fish farms. Theme name
Live fish and egg movements
Exposure via water
On-site processinga
Short distance mechanical transmission
Distance independent mechanical transmission
Theme definition
Introduction of pathogen through consignments of infected live fish (for ongrowing, stock enhancement or processing) and contaminated eggs from other sites, but not including pathogen present in transport water or contaminated transport tanks, lorries nets or packaging
Direct introduction of pathogen to farm source water by susceptible fish populations and activities upstream: i.e. other farmed and wild stocks (including fish released by farms for restocking or conservation purposes); fish processing
Introduction of pathogen through infected dead fish sourced from other sites for on-farm processing
Introduction of pathogen from sources in close proximity to the farm through routes including: piscivorous birds or other animals
Introduction of pathogen via contaminated fomites including: fish transporters and associated equipment; other equipment or personnel shared with other fish farms; fishery activities; visitors; vehicles, etc.
a
ISA experts were not consulted on theme ‘on-site processing’.
assume that all pathways were present on all farms. In scenarios 3 and 4 it was made clear that the farm sites were in approved disease-free zones. Part 2 of the questionnaire asked the experts to assume scenario 1 and allocate weights and probabilities for pathways within the 5 risk themes. Theme ‘processing on-site’ was omitted from the ISA questionnaire, since it was considered not relevant due to current management practices and regulations in Europe. Because of the context provided to the experts in our questionnaire, the VHSV genotype consulted on in our study was VHSV genotype Ia. The experts participating in the consultation are described in Tables 2a and 2b. Experts were mainly drawn from national fish health services for participating countries. Initial identification of experts was through personal contacts of the principal author (BO) and based on knowledge of experience in investigating outbreaks of notifiable fish diseases (field investigations or data analysis) or on their comprehensive research experience. Because EHN has not been reported in the EU, no expert consultation was undertaken for this disease. Questions were included in the introductory section of the questionnaire to assess the pathogen-specific expertise of the participants. One of the responses to the ISA online questionnaire was based on the agreed answers of 3 experts working in the same organisation. Four of the KHV experts were from the same organisation; their responses to the online questionnaire were combined into a single response (calculated mean).
2.1. Expert consultation meeting Experts who had participated in the online questionnaire were invited to a 2 day consultation meeting. Where multiple experts from a single organisation had contributed to the online questionnaire for a specific disease, responses had been combined into a single value, and only one expert per organisation and per disease was invited to the physical meeting to avoid bias. This meeting was structured as: (1) Group discussion I, (2) Plenary discussion I; (3) Group discussion II, and (4) Plenary discussion II.
Table 2a Expert consultation on risk factors for introduction of infectious pathogens into fish farms. Information about experts participating in the online questionnaire.
Total number of experts responding to online questionnaire Employment status Employed by non-government organisation Self-employed Employed by government organisation Describe your current work: • Analysis of fish farm production data • Collection of fish farm production data • Design of surveillance programmes for fish diseases • Implementation of fish health regulations/legislation • Analysis of data/information on the occurrence of disease in fish farms • Investigation of fish kills in wild populations • Laboratory diagnosis of fish disease • Aquatic animal health research • Collection of data/information on the occurrence of disease in fish farms • Advice to government on aquatic animal health • Investigation of disease outbreaks on fish farms
Number of experts
% of total
22
100
3
14
4 15
18 68
6
27
9
41
9
41
9
41
10
45
11
50
11
50
12
55
14
64
15
68
18
82
Frequent visits to fish farms? (more than 4 times per year) 9 No 13 Yes
41 59
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
G Model PREVET-3549; No. of Pages 17
ARTICLE IN PRESS B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
4
Table 2b Expert consultation on risk factors for introduction of infectious pathogens into fish farms. Responses to online questionnaire, country and experience of participating experts; expert attendance during workshop. Some experts were invited to respond to online questionnaires for more than 1 disease. VHVS Online questionnaire 15 • Experts invited to participate in online questionnaire (n) 12 • Responses (n) • Main place of work (country) of experts completing online questionnaires (n) Austria 2 Czech Republic 1 Denmark 1 England 1 Faroe Islands 0 Germany 2 Italy 1 Norway 0 Poland 1 Scotland 0 Switzerland 3 • Experience of experts responding to online questionnaire Years worked in aquatic animal health Mean 19.6 18 Median Min 2 Max 35 Number of investigations into outbreaks of the disease involved in over the past 10 years 25.6 Mean 26.5 Median 1 Min 60 Max Workshop • Experts attending workshop (n) • Experts in working group (n) a b
10 9
IHNV
KHV
ISAV
15 11
12 11 (8)a
8 5b
2 1 0 1 0 2 1 0 1 0 3
2 1 0 5 (2)a 0 2 0 0 1 0 0
0 0 0 0 2 0 0 4 (2)b 0 1 0
18.2 16 2 35
20.6 20 8 39
15.2 12 8 25
8.0 6 0 30
35.0 25 8 95
9.9 10 1 34
10 9
7 4
3 3
Responses from 4 experts combined to a single response (arithmetic mean). Three experts provided a single joint response and counted as 1 response.
Group discussion I: Experts were split into pathogenspecific break-out groups chaired by a facilitator. The concept behind the 4 scenarios in the questionnaires was introduced by the facilitator and clarification provided until the experts were confident that they understood the assumptions. The combined result of the responses from all experts for a given question from the online questionnaire was presented to the group in a boxplot. The experts were thus given the opportunity to consider the others’ opinions with the objective to agree on a consensus value from the whole expert group for each parameter (Delphi approach). The facilitator managed the discussion by: (1) identifying consensus when reached; (2) highlighting conflicting views; (3) finding compromise and (4) encouraging contributions from all group members. Plenary discussion I: The results from the group discussions were presented to all participants followed by discussion. Group discussion II: Previously compiled summaries of published information, relevant to assessing the scores for the various risk factors and themes for each pathogen were presented alongside the outputs from Group I discussions and experts were given the opportunity to revise scores. Plenary discussion II: Outcomes of the second group discussions were reported back to all participants. Experts from all groups therefore had an opportunity to comment, and query the results of other breakout groups. The detailed outcomes of the sessions and discussions were documented (available from the corresponding author on
request). The output included: agreed most likely, minimum and maximum values (single values agreed by the expert groups) for each of the parameters, with explanatory information where needed, such as identification of factors that may influence the parameter. 3. Results 3.1. Online questionnaire and expert consultation meeting On average, the experts had more than 15 years experience working in the field of aquatic animal health and, for those diseases widespread throughout Europe (VHS, IHN and KHV), most experts had investigated a substantial number of outbreaks of the disease for which they acted as experts (Tables 2a and 2b). All experts’ responses were used to calculate parameter estimates. 3.2. Parameter estimates obtained Results (median, minimum and maximum weights and likelihood estimates) from the expert elicitation are shown in Tables 3–5 (abbreviated questions; the full questions appear in Appendix 1). The weight medians do not always sum to 100 as they were calculated separately for each risk theme based on the responses provided by the experts. Some experts provided estimates in the initial online questionnaire that were clearly discordant with
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
Scenario 1
Risk theme
LFM
Water
Scenario 2
Pathogen VHSV Median Min Max
70 10 88
IHNV Median Min Max
Scenario 3
Scenario 4
Short
D-in
LFM
Water
Proc
Short
D-in
LFM
Water
Proc
Short
D-in
LFM
Water
Proc
Short
D-in
10 2 20
5 1 20
2.5 0 20
8 2 20
68 20 80
13 4 55
4 1 20
3.5 1 21
9 1 25
50 5 96
14 0 17
4 0 15
6 0 20
14 4 40
40 5 96
16 0 69
7.5 0 20
7.5 0 20
13 2 40
80 40 90
9 2 20
3 1 20
2 0 20
10 1 35
80 40 85
9 2 21
2 1 15
2 1 20
10 2 25
50 5 96
10 0 50
2 0 15
4 0 15
10 3 40
50 5 94
10 0 50
1 0 20
5 0 20
10 4 40
KHV Median Min Max
73 60 95
12 1 27
2 0 8
3 1 7
10 2 12
67 55 95
17 1 30
2 0 5
5 1 7
8 2 16
65 20 78
5 2 25
4 0 15
6 1 10
20 3 44
60 10 78
5 2 24
4 0 10
12 2 20
19 3 44
ISAV Median Min Max
5 1 9
50 40 60
2 1 3
42 15 40
8 3 13
45 35 55
2 1 3
45 40 50
60 10 99
20 10 25
20 0.3 40
60 10 99
20 10 25
0 0 0
0 0 0
20 0.3 40
Weights for risk themes assuming scenario 1–4. Scenario 1: 2% between farm site-level prevalence across the whole country; scenario 2: 5% between farm site-level prevalence across the whole country; scenarios 3 and 4: assumed approved disease-free zones within the country, but outside these zones between farm site-level prevalence of 2 or 5%, respectively. Pathogens: VHSV, viral haemorrhagic septicaemia virus; IHNV, infectious haematopoietic necrosis virus; KHV, koi herpes virus; ISAV, infectious salmon anaemia virus. Risk themes: LFM, live fish and egg movements; Water, exposure via water; Proc, on-site processing; Short, short distance mechanical transmission; D-in, distance independent mechanical transmission.
Table 4 Expert consultation on risk factors for introduction of infectious pathogens into fish farms. Estimates of likelihood of transmission of pathogen with live fish movements from a subclinically infected source site to a receiving rainbow trout (VHS, IHN), carp (KHV) or Atlantic salmon (ISA) farm site, by species. IHNV
VHSV
KHV
ISAV
Species
Median
Min
Max
Species
Median
Min
Max
Species
Median
Min
Max
Species
Median
Min
Max
Rainbow trout Brown Trout Brook Trout Grayling Pike
80 30 7.5 37.5 0.5
10 1 0 0 0
100 85 10 60 1
Rainbow trout Atlantic salmon Brown Trout Brook Trout Pike
80 60 5 5 5
10 2 0 0 0
95 95 10 10 10
Common Carp Goldfish Grass Carp Tench
70 20 20 10
30 2 0.75 0.75
90 40 20 20
Atlantic salmon Rainbow trout
95 10
90 5
100 15
ARTICLE IN PRESS
Proc
G Model
PREVET-3549; No. of Pages 17
Scenario
B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
Table 3 Expert consultation on risk factors for introduction of infectious pathogens into fish farms.
5
G Model PREVET-3549; No. of Pages 17
6
ARTICLE IN PRESS B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
the majority. Reasons for such deviations were clarified in all cases by discussion (misinterpretation of the question or regional differences). Following clarification, the group estimates were agreed by all experts in the specific working groups and approved by the plenary. No changes were made by any of the groups based on reflection of the published information (Group discussion II). Median estimates are used in the sections below unless stated otherwise.
3.2.1. Weights for themes (Table 3) For VHSV, IHNV, and KHV, live fish movements was the most important risk theme for all scenarios (Table 3). Live fish movements were considered less relevant for farm sites located in zones with Category I status (approved pathogen-free) in scenarios 3 and 4, compared to farm sites located in an endemically infected country, although it remained the most important risk theme. Experts assumed compliance with current regulations, i.e. farmers in Category I zones would source fish from farms of equivalent health status. With higher assumed prevalence (scenario 2 and 4 vs. scenario 1 and 3 respectively), the scores for live fish movements decreased, while they increased for transmission via water or short distance mechanical transmission (VHSV and KHV). The range of values (minimum to maximum) for some themes and diseases was both wide and skewed (e.g. live fish movement for scenario 1 VHS has a median of 80, a minimum value of 10, and a maximum of 88). For the seawater disease pathogen (ISAV), exposure via water and distance-independent mechanical transmission were the most important themes for scenarios 1 and 2. Live fish movements were rated the most important risk theme for scenarios 3 and 4. Farm sites located in Category I zones were considered to be at a lower risk via mechanical transmission and water (compared to farms located in an endemically infected country – scenarios 1 and 2), which increased the importance of other themes, notably live fish movements. The minimum to maximum ranges were lower than for VHSV, IHNV and KHV.
3.2.2. Likelihood of transmission of pathogen with live fish movements (Table 4) The experts were asked to estimate the likelihood of transmitting pathogen through live fish movements from a subclinically infected source site by different species. The likelihood estimates (medians) provided for a single movement ranged from 70% (carp, KHV) to 95% (Atlantic salmon, ISAV) (Table 4). The ranges (min to max) around these median values were very wide for VHSV, IHNV and KHV (e.g. 10–100 for VHSV transmission by rainbow trout) but narrow for ISAV transmission by Atlantic salmon (90–100). For VHSV, IHNV and KHV, experts were asked to consider species not currently listed as susceptible to these diseases by the Directive (i.e. brook trout for VHSV; brown trout, brook trout and pike for IHNV; goldfish, grass carp, and tench for KHV). The medians for these species ranged from 5 to 20.
3.2.3. Scores for pathways within the risk themes (Table 5) The results presented in Table 5 cover questions 26–44 of the VHSV questionnaire (detailed estimates for specific pathways within the 5 risk themes) and the corresponding questions within the questionnaires for the other pathogens. The experts were asked to imagine a hypothetical country with 2000 farm sites and 2% prevalence between farm sites for each specific pathogen. Under the assumption that the source site was subclinically infected, the experts estimated that of 100 consignments of non-disinfected rainbow trout eggs, 40 would lead to infections of VHSV and 30 of IHNV at receiving sites. Of 100 consignments of Atlantic salmon eggs, 1 would lead to infection of ISAV at receiving sites. If eggs were disinfected, the experts reduced these estimates to 1 (VHSV), 5 (IHNV) or 0.01 (ISAV) (Q26). This question was omitted for KHV, since transport of fish eggs between sites is not common practice for cyprinid aquaculture. Within theme ‘exposure via water’, presence of farms holding species susceptible to infection with the specific pathogen within 5 km upstream/sea-distance was considered the highest risk (Q30). Estimates were that over a 1 year period, out of 100 independent farm sites (located in independent water catchments/sea water zones), there would be 32.5 infections of VHSV, 20 of IHNV, 30 of KHV or 20 of ISAV if an infected farm was located within 5 km upstream/seaway distance (Q27). The number of farm sites estimated to become infected over a 1 year period as a result of the presence of infected wild fish within 5 km of these farm sites was 10 (VHSV and IHNV), 15 (KHV) and 5 (ISAV) (Q28). For VHSV and IHNV, the risk of infection from the presence of a fish processing facility within 5 km upstream was rated similar to the presence of wild fish populations or stocked fisheries (susceptible species) within 5 km upstream (Q30). On-site processing was considered an important risk for VHSV, IHNV, and KHV introduction (Q32–36). The experts estimated that out of 100 farm sites receiving rainbow trout (or carp) carcases for processing from other farm sites (of unknown infection status), 5 would become infected over a 12-month period (VHSV, IHNV, KHV) (Q33). When the processing site received live infected fish, the estimates increased to 77.5 (VHSV), 77 (IHNV) and 80 (KHV) (Q36). The risk of infection due to short distance mechanical transmission (assuming no direct water connectivity) was estimated to be less than 6 of 100 farms per year across all scenarios and diseases (Q37). The highest estimates were for KHV, 4.6 (scenario I), 5.8 (scenario 2), 2.5 (scenario 3) and 3.8 (scenario 4) of 100 farms, compared to 3 or fewer sites for VHSV, and 2 or fewer for IHNV. The higher the prevalence in the country, the higher the number of farm sites expected to become infected. The experts for ISAV estimated that no farm sites in a declared disease-free zone would become infected due to short distance mechanical transmission, and 0.5 or less of 100 farm sites to become infected in an endemically infected country outside a disease-free zone. In ranking three potential sources for infection via this route (Q38), fish farms tended to be considered the most relevant, then fisheries
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
Q
Topic of question
VHSV Median
IHNV Min
Max
Median
KHV Min
Max
Median
ISAV Min
Max
26
Eggs not disinfected Eggs disinfected
40 1
0 0
100 5
30 5
1 0
98 10
N/A N/A
Exposure via water
27 28 29
Farm Wild fish Fisherya
5 0.5 1
100 50 100
80 20 80
80 50 60
Farm Processor
30 15 30 N/A 50 5
1 2 15
Transport route (ISA) Weight by source
20 10 12 N/A 50 20
1 0.5 1
30
32.5 10 20 N/A 50 25
20 0
80 20
Wild/fisheryb Transport route (ISA) Other
5
60
0
20 55 100 10 70 50 95
On-site processing
Short distance mechanical transmission
Distance independent mechanical transmission
a
c d e f
17.5 N/A 5
5
45
0
10
40 47.5 5 27.5 10 77.5
1 3 0 5 1 10
80 93 50 100 50 100
Scenario 1 Scenario 2 Scenario 3 Scenario 4 Farms Fisheries Wild 5 hae
33 39
Within 2 km
Min 0 0
Max 2 0.02
20 5 10 5 79 5
5 1 2 2 20 2
40 10 20 10 100 10
10 5 1
2 2 0
20 10 2
40 5
75 42
5
45
0
10
36 N/A 9
30 49 5 20 10 77
1 3 1 5 1 15
90 99 10 90 50 99
40 60 5 15 20 80
0 45 0 3 1 10
5 20 5 3 75 40 40 60
1 2 0.5 1 60 30 10 33
0 0 0 0 20 20 1 5
5 4 2 3 70 40 40 60
4.6 5.8 2.5 3.8 40 40 20 28.3
0.1 0.2 0.1 0.1 20 20 2 10
13.7 20.5 27.4 38.4 85 75 40 60
0.1 0.5 0 0 95 N/A 5 N/A
0 0 0 0 90 2
10
20 10
45 75
33 33
20 10
45 75
35 35.9
30 10
45 60
40 60
30 50
50 75
60 25 15
40 5 1
94 35 30
65 25 10
40 5 1
94 35 30
60 27.5 10
20 20 2
85 45.8 40
75 20 5
60 10 2
80 40 10
19 N/A 1
Process No processing Dead unknown Dead infected Live unknown Live infected
37
No. farms infected
38
Weight by source
39
Proportion farms inf.
40
Fishery/farmf
41
Scenario 1 Scenario 2 Scenario 3 Scenario 4 Staff Equipment Fish vehicles General vehicles Fishery Unauth people Unauth vehicles Insufficient dis vehicles Insufficient dis footwear Staff
2.5 3.5 1 2 20 35 22.5 5 10 7.5 5 30 30 30
1 1 0 0 5 5 8 0 2 0 0 10 10 5
10 10 3 5 30 50 50 20 30 15 10 40 40 70
2 4 1 2.5 20 35 25 5 10 5 5 30 30 30
1 1 0 0.5 5 5 8 0 2 0 0 10 10 5
10 30 5 5 30 40 50 10 25 15 15 40 40 70
7.3 13.8 4.7 5.4 12.5 37.5 20 1 12.5 5.6 6.9 30 30 30
1 1 1 2 3 15 5 0 5 1 0 0 5 5
21 30 29 29 30 45 40 10 75 15 10 60 35 93
2 10 1 5 2 30 60 8 N/A 1 1 80 1 17
1 2 0 2 1 20 50 2
4 20 2 10 4 45 75 20
0 0 75 0 10
2 2 95 2 25
Avian Mammalian Stored waste Staff
20 7.5 40 22.5
1 1 20 1
50 30 80 65
20 5 40 26
1 1 20 5
50 30 80 65
20 10 40 24.9
2 2 20 5
45 30 70 60
4 1 80 15
1 0 70 5
8 2 95 30
43
b
75 45
1 0.01
33 34 35 36
32
42
Short distance and distance independent mechanical transmission
30 5
Median
44
0.2 1 0 0 100
For ISAV the experts provided an estimate for presence of processing site within 5 km. For ISAV: the experts provided an estimate for presence of wild fish only within 5 km. For ISAV: proportion of farms with 1,000,000 fish on site. For ISAV: the experts provided an estimate for presence of farm sites within 5 km.
ARTICLE IN PRESS
Live fish and egg movements
G Model
PREVET-3549; No. of Pages 17
Theme
B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
Table 5 Expert consultation on risk factors for introduction of infectious pathogens into fish farms. Relative weights or likelihood estimates for pathogen transmission for individual risk pathways within risk themes. The number in column Q refers to the question no. in the VHS questionnaire. A keyword indicates the subject of the question.
7
G Model
ARTICLE IN PRESS
PREVET-3549; No. of Pages 17
B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
8
Table 6 Expert consultation on risk factors for introduction of infectious pathogens into fish farms. Q
VHSV
IHNV
KHV
ISAV 1 15 30 4
42
Staff Equipment Fish vehicles General vehicles Fishery
1 1.75 1.13 0.25 0.5
1 1.75 1.25 0.25 0.5
1 3 1.6 0.08 1
43
Unauthorised people Unauthorised vehicles Insufficiently disinfected vehicles Insufficient disinfected footwear Staff
0.25 0.17 1
0.17 0.17 1
0.19 0.23 1
0.06 0.06 4.71
1
1
1
0.06
1
1
1
1
Avian Mammalian Stored waste Staff
0.89 0.33 1.78 1
0.77 0.19 1.54 1
0.80 0.40 1.61 1
0.27 0.07 5.33 1
44
Relative importance of mechanical transmission pathways of VHSV, IHNV, KHV and ISAV. Medians of estimates provided by experts to questions 42–44 on individual mechanical transmission pathways scaled against transmission via staff. The number in column Q refers to the question no. in the VHS questionnaire. A keyword indicates the mechanical transmission pathway.
and thirdly wild fish populations. Farm size was considered to have relatively little impact on the likelihood of a farm becoming infected through short distance mechanical transmission (Q39). Distance to other farms or fisheries was highly relevant (Q40). For distance independent mechanical transmission the estimates were consistently highest for KHV across all scenarios and highest for scenario 2 across all diseases (Q41). Mechanical transmission routes were compared in three separate questions (Q42–44). Risk of transmission via staff was included in all three so that responses could be scaled against this pathway for each question. Shared equipment and storing fish waste on site were identified as the highest risks for mechanical transmission of VHSV and IHNV. Vehicles were considered the most relevant route for ISAV, and equipment most relevant for KHV (Table 6). Details of adjusted most likely values and discounted outliers emerging from the expert consultation meeting compared to the online questionnaire results are available from the corresponding author on request. 4. Discussion The use of expert opinion is an accepted approach where information is not available (Gustafson et al., 2013). Expert elicitation has been used in the context of diseases in terrestrial animals and also for fish diseases (Bruneau et al., 1999; Gustafson et al., 2005, 2010, 2013). We used a twostep approach based on Delphi technique (Dalkey, 1969). A facilitator managed discussions with defined instructions to ensure participation from all experts, previously recognised as a potential issue (Oidtmann et al., 2011a). Experts were given four scenarios to explore the importance of disease prevalence and status. An initial setting in which the country was already infected was chosen to avoid a focus on pathways of introduction into areas
previously free of the disease. A prevalence of 2% was chosen, because it is a commonly used design prevalence for the first level of clustering (e.g. proportion of infected farm sites in a zone) for animal diseases when demonstrating freedom from disease (Corsin et al., 2009; World Organisation for Animal Health OIE, 2010). Thus scenario 1 best reflects the potential situation in a country that wishes to demonstrate freedom from infection. Scenarios 3 and 4 were used to identify whether there were differences in importance of the routes depending on whether a farm site was located in a country aiming to achieve disease-free status opposed to a farm site in a zone. The format of the questionnaire was novel to the majority of experts and some found it difficult to understand the questions asked. However, following the expert group work, there was a consensus amongst all experts that the values agreed during the group sessions were appropriate. In all groups, consensus was achieved regarding the most likely values for all parameters. Consensus was assessed by inviting experts individually to comment on the most likely values and the range (minimum and maximum values) for risk pathways in the Group discussions. Particular attention was paid to obtaining comments from experts who were less active in group discussions. Variability in the responses to the online questionnaire (minimum and maximum values, Tables 3–5) usually reflected the different context and industry structure in the countries represented. Experts agreed during the meeting that the values provided in the online questionnaire were suited to describe the range these values could take (small modifications were made during Group discussions, where an outlier stemmed from a misunderstanding of a question). For the remote part of the expert consultation (the online questionnaire), individual experts had not been asked to quantify a range around an estimate. Ranges were obtained by summarising the responses from all experts for a specific pathogen to a given question. Given the paucity of information and the widely differing production systems for the same species across the member states represented, it is not surprising that the estimates of the weights and likelihood estimates had wide ranges. During discussions in the meeting it emerged that some of the variation in experts’ scores is explained by the variation in level of exposure to risk pathways between regions. However, the experts agreed that the medians were the appropriate most likely values and that the lowest and highest estimates provided by experts in the online questionnaire suitably represented the appropriate range the respective estimates could take. Several of the experts’ estimates, such as likelihood of transmission of virus via subclinically infected fish, are independent of regional practices and therefore have relevance across the board for the given pathogen. Risk of pathogen introduction to a farm site is dependent on whether a certain exposure pathway applies and its frequency. Therefore, identification of farms at higher risk of becoming infected with a specific pathogen is ideally informed by the presence or absence of certain risk pathways and the level of exposure to this pathway if it applies. Formulas can be devised to take level of exposure into account (e.g. number of live fish introductions to a given farm site; number of movements of vehicles or equipment
Please cite this article in press as: Oidtmann, B.C., et al., Expert consultation on risk factors for introduction of infectious pathogens into fish farms. PREVET (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.03.017
G Model PREVET-3549; No. of Pages 17
ARTICLE IN PRESS B.C. Oidtmann et al. / Preventive Veterinary Medicine xxx (2014) xxx–xxx
– with/without disinfection – onto a site, etc.). However, there may also be scope for further narrowing the ranges appropriate for a certain region through further expert consultation using local experts. The approach used in this expert consultation followed several recommendations recently summarised by Gustafson et al. (2013): experts were individually contacted and the questionnaire discussed with them before they were asked to respond, which allowed minor adjustments to the questionnaire or further explanations. Leading questions were avoided (with the intention of avoiding anchoring) and counts relating to clearly specified denominators were elicited rather than probabilities. Experts then met for group discussions to share the results of the online questionnaire and discuss the distribution of the responses. Weighting of responses has been suggested to reflect different levels of experience of participants in an expert consultation (Cooke, 1991; Cooke and Goossens, 2000; Aspinall, 2010). However, peer and self-evaluation may be unreliable predictors of expert performance. More accurate estimates from expert consultations can result from systems that provide feedback about experts’ performance, encourage experts to think about their estimates and aggregate individual estimates (Cooke, 1991; Fisher, 2009; Burgman et al., 2011). In fact, expert advice will be more accurate if based on broadly defined expert groups, structured question protocols and feedback (Burgman et al., 2011). These principles we have tried to apply in this study. 4.1. Relevance of live fish movements and exposure via water
9
diseases (e.g. brook trout (Salvelinus fontinalis) for both VHS and IHN) were nevertheless considered to carry a significant risk for pathogen transmission. 4.3. On-site processing On-site processing received a relatively low weight (
