The Scottish Government

« Previous | Contents | Next »

Listen

4. METHODS AND RESULTS

4.1 Description of methodological approaches

Our methodological approaches have been divided between economic methods, incursion scenario methods, epidemiological scenario methods and Scottish vector data methods to ease rapid assimilation. Clearly there had to be considerable overlap between the teams.

Incursion scenarios methods:

Three main routes for potential incursions were identified with agreement from SG and a number of approaches were used to determine which of these potential routes posed the greatest level of risk.

1. Wind-borne dispersal of vectors from south-east England, Northern Ireland or continental Europe: The risk of incursion via wind-borne midges was assessed using ten years worth (1998-2007) of data on wind speed and direction and temperature. These were used to determine the frequency of winds suitable for carrying vectors from potentially infected areas to Scotland.
2. Import of infected animals: The risk of introduction via the import of infected animals was examined using movements data for 2006 to provide the number of movements to each Scottish county by month.
3. Northwards spread of BTV from south-east England: The risk of northwards spread was investigated using a model for the transmission of BTV between farms (see ANNEX 2 (a)). This was used to predict if and when BTV is likely to arrive in Scotland, following expansion from the current infected area in south-east England, assuming that only minimal control measures were applied. Analysis of climatological data (see (i) above) was also used to assess the risk of incursion if disease foci were to arise near the Scottish border.

More detail on the risk of incursion was added to the analyses by using the relationship between temperature and the extrinsic incubation period ( EIP) to predict when and where vectors are likely to pose a transmission risk. This was done by linking an accumulated degree-hour model for the completion of the EIP with temperature data for Scotland.

Epidemiological scenarios methods:

For each of the incursion scenarios considered, the impact of a number of control strategies was investigated:

1. Implementing only the minimal requirements;
2. Vaccinating 100% of holdings in a border protection zone ( PZ) (see ANNEX 3 (a)).
3. Vaccinating 80% of holdings in a PZ to the Highland B/F line (see ANNEX 3 (a));
4. Vaccinating 50% of holdings in a PZ comprising the whole of Scotland ¹; and
5. Vaccinating 80% of holding in a 100km PZ around the first identified holding (only applied if the incursion occurs above the Highland B/F line).

For incursions which occur in 2008, vaccination strategies were reactive, whereas for incursion occurring in 2009, they were prophylactic with vaccination taking place in January 2009. In control scenarios 2-5, additional reactive vaccination was applied (at 100% uptake) in a 20km control zone around any infected holding. A total of 21 incursion/control scenarios were considered (see Table in ANNEX 3 (b)).

The spread of BTV under each incursion/control scenario was assessed using a stochastic, spatial model for the transmission of BTV in Scotland (see ANNEX 2 (a) for a description of the model, including underlying assumptions and parameter estimation). For each scenario 100 replicates of the model were simulated with the initial conditions specified according to the incursion scenario. Importantly, only a single incursion event was considered. Each replicate was run for two years, starting in January of the year in which the incursion occurred.

Scottish vector data methods:

Candidate midge vectors for BTV in Scotland include farm-associated members of the Culicoides pulicaris and Culicoides obsoletus complexes. These have been incriminated by fine-scale overlap of their distributions with outbreaks, by isolation of virus from wild-caught adults in several sites across Europe and by vector competence experiments on UK populations (Carpenter et al. 2006). The potential role of the Scottish biting midge, C. impunctatus (also a member of the C. pulicaris complex) is more difficult to ascertain since this species generally prefers to bite humans and is autogenous - meaning that it does not require a blood meal to lay its first egg batch as an adult. For transmission to occur an infected vector must take a minimum of two blood meals, the first to acquire an infection and the second to pass the infection to a new host. In between these meals, virus must have replicated inside the midge and spread to the salivary glands and the time for this extrinsic incubation period depends on temperature. Since the interval between meals depends on the length of the reproductive cycle, the likelihood of transmission is very sensitive to the relative timing of the reproductive cycle and the extrinsic incubation period. Since C. impunctatus does not feed before the first egg batch, this species must survive to complete a minimum of three reproductive cycles (each taking between 5 to 8 days with around 60% of females surviving each cycle) before transmission can occur. Most midge vectors that feed before every egg batch must complete a minimum of two reproductive cycles to transmit virus.

Though UKC. impunctatus populations have been found to have relatively low levels of competence for BTV in the laboratory (~ 0.1-0.2%, Carpenter et al. 2006, Jennings & Mellor, 1988), vector species with low competence can still play a large role in transmission if they are highly abundant (e.g. C. variipennis sonorensis in North America). C. impunctatus is enormously abundant and widespread across Scotland particularly in bog/heathland areas and in the Highlands. Given the overlap of this species with both farm-associated vectors and wild ruminants, it cannot be discounted as a potential vector of BTV or other midge-borne pathogens. Wild ruminant species are considered as a potential reservoir for BTV, with both white-tailed deer ( Odocoileus virginianus) and pronghorn ( Antilocapra americana) being susceptible to disease in the US. During the recent BTV-8 outbreaks, fallow deer ( Dama dama), roe deer ( Capreolus capreolus), mouflon ( Ovis mouflon - a wild sheep species) have all tested positive for BTV-8 in Germany ( VLA, 2007) - albeit at low seroprevalence in an area with high prevalence in domestic ruminants.

In Scotland, the offspring of sheep bred in the Highlands, alongside wild ruminants (including roe, red, sika and fallow deer) and large C. impunctatus populations in summer, are brought to the lowlands areas in autumn, to live alongside cattle and farmland midge species until spring. These practices may provide potentially frequent opportunities for BTV to be transferred between farm-associated and bog/heathland vectors and between wildlife reservoirs and domestic ruminants. Potential interactions between hosts and vectors for bluetongue in the Scottish landscape can be summarised then according to the schematic below ( Table. 1).

To describe, explain and predict geographic variation in the abundance of different vector species, one would ideally be armed with detailed knowledge of each species' competence levels and breeding site requirements as well as rich seasonal demographic data from many locations (with which to parameterise environmentally-driven statistical and biological models). Such data are largely lacking at present, but are being gathered for all potential vector groups across Scotland as of late 2007 by a parallel RERAD project (lead by APS Ltd).

Table 1 Schematic of the vector and host communities for BTV across Scotland

This project has provided us with a limited dataset on autumn vector abundance for this study. Given the 'data gaps' above our sub-project objectives were constrained to be the following:

1. Relate autumn abundance of farm-associated midge vectors across Scotland to habitat and micro-climate variables (within statistical abundance models) with a view to producing predictive maps that indicate approximate levels of abundance of each of the important vector species. Such maps would enable us to estimate variation across Scotland in the ratio of vector to hosts - an important 'ingredient' of R ₀ transmission models for vector-borne diseases.
2. .Culicoides impunctatusMap habitat for the Scottish biting midge, Since C. impunctatus is a major biting nuisance to humans, the habitat preferences and seasonality of this species have been relatively well-studied in Scotland. We aimed to extract characteristics of this species' preferred habitat from literature and expert knowledge and to map, qualitatively, the extent of this habitat in relation to ruminant densities and farmland across Scotland.
3. Map densities of susceptible hosts including wild ruminants and likely areas of interactions between hosts

Economic methods:

The basic model used for estimating the costs and benefits of BTV incursion and control freedom has been previously used for calculating the direct costs associated with endemic diseases of livestock in Great Britain (Bennett et al., 1999). This spreadsheet model was based on the risk of livestock contracting a disease and associated costs of prevention, treatment and reduced performance. Menzies et al. (2002) applied this methodology to estimate the direct costs of cataracts in farmed Norwegian salmon. A spreadsheet model similar to that of Bennett et al. (1999) and Menzies et al. (2002) was adapted and extended by Moran and Fofana (2007) to account for the cost and benefits of fish disease incursion and control in the UK. The spreadsheet model methodology developed by Moran and Fofana (2007) was applied here to gather all data from the output of the BTV8 epidemiological model, trade, control, monitoring and surveillance costs of animal diseases in Scotland.

In economic terms, a cost avoided from an action is a benefit of that action. In the case of animal disease, the benefits of measures to prevent or reduce the deleterious effects of disease on animals include avoiding costs from the effects of disease, which would otherwise have occurred (Malcolm 2003). The benefits of avoiding BTV incursion in Scotland include both the output losses and control costs (e.g. vaccination costs and movement restrictions) of dealing with an incursion. These were termed the total cost of disease at farm level by McInerney (1996). It is important to appreciate that these total costs are part of the 'benefits' and not the 'costs' in the following cost benefit analysis. Further explanation of this issue is therefore given in ANNEX 1h. Our approach to the benefit-cost comparison was to consider the avoided costs of an outbreak as the benefits accruing to surveillance outlays i.e. those expenditures both public and private that are incurred in the hope of avoiding a BTV incursion or reducing its severity. These outlays are the unavoidable costs as they occur whether or not an incursion of BTV occurs in Scotland and are effectively constant across all incursion scenarios investigated. As such they form a benchmark against which to judge the impact of the different incursion scenarios and the alternative controls applied in each case. They also represent the current situation, i.e. emphasis on maintaining freedom from BTV incursion into Scotland.

An insight into the comparison of benefit-cost is to consider a scenario where there is no or limited disease surveillance. Any outbreak (ignoring for now risk and/or frequency of outbreak) incurs a total cost of C ₁, consisting of a range of damage impacts across the industry. The implementation of an improved surveillance programme entails eventual outbreak cost of C ₂; (C ₂ < C ₁). Any outbreak that occurs with this in place is necessarily identified and curtailed more rapidly. This alternative regime could be one of any number (say n) of configurations of surveillance that mix voluntary and mandatory measures. Surveillance benefits are then denoted as B = (C ₁ - C ₂): the difference in costs in terms of the severity of the damage of the outbreak as a result of having surveillance in place. If the probability of incursion (and thus total costs C ₁of BTV are incurred) in any year is R, then the expected benefit in any year as a result of having the surveillance is an expectation R • B.

The costs of implementing surveillance programmes were C _{3 ².} With this information a net-benefit estimate (R x B - C ₃₎ could be derived for the surveillance option, or a benefit-cost ratio for the programme (R x B/ C ₃). If there is no estimate of probability of incursion, as is often the case for ex-ante evaluations, the benefit-cost ratio can be simply evaluated as B/C ₃.

The appraisal of policy options using CBA require the identification of a baseline or status quo scenario against which the costs and benefits of alternative government interventions are evaluated. C ₃ in the example provided above can be regarded as the baseline. It is the status quo of the investment which the public and private sectors make in the implementation of surveillance programmes to prevent BTV incursion or to limit the deleterious effects of an outbreak. In order to derive the cost related to investment, a percentage of public sector disease surveillance and control expenditure was assumed to be passively dedicated to BTV. In the private sector, good animal husbandry practices and expenditure on veterinary services are usually not disease specific but are meant to keep any form of disease at bay thus passively limiting the deleterious effects of BTV. The private sector cost estimated in this category was added to the estimated public sector cost assumed to be dedicated to surveillance and control of BTV to provide the total baseline or status-quo cost (see ANNEX 1a for specific assumptions). It was against this cost that all made-up or counterfactual scenarios (see ANNEX 1b for matrix of counterfactual scenarios) costs and benefits of alternative intervention are evaluated.

CBA requires the identification of the full range of costs and benefits associated with animal health surveillance policy actions and their measurement in physical terms. This entails the understanding, measurement, explanation and/or prediction of the impacts of BTV. This stage of the analysis is easier in ex-post evaluations and extremely complicated in ex-anteCBA due to the risks and uncertainties attached to making forecasts and predictions in this field. In this case, estimating the benefits of avoiding disease with (C ₂) and without (C ₁) investment in surveillance programmes (C ₃) was particularly difficult due to lack of information about the effects of surveillance on the nature and extent of BTV incursions and on the probability of specific BTV incursions (R). We therefore adopted a 'null hypothesis' that surveillance would be successful i.e. that C ₂ = 0 and thus the benefits of surveillance B= C ₁ i.e. the total costs of the avoided incursion that would otherwise ensue without surveillance. This overstates the benefits but equally across all incursion scenarios and disease control options. It therefore does not affect the ranking of the alternative disease control (vaccination) strategies assessed.

All costs of a disease (whether baseline costs of prevention or their benefits due to disease losses avoided) are generally categorised into direct and indirect costs. Direct costs of diseases are the losses that may occur at farm input and output levels. At the input level, the costs are attributable to losses when disease destroys the basic resources of the livestock production process. The total direct cost of a disease is the sum of the production losses (direct and consequential) and the costs of disease control. Consequential on-farm losses include losses due to the fall in stock numbers, restrictions of movement when zoning restrictions are put in place and due to the loss in animal value.

Indirect costs are costs associated with revenue forgone through loss of markets, sub-optimal production methods and additional costs incurred to eradicate diseases. These costs were estimated by simulating the potential impacts on livestock and livestock product prices along the value chains. It was not possible to account for all that should be included in this category due to the methodological difficulties and constraints of this type of analysis and the extensive data requirements. The costs included are the business disruption costs suffered by farmers, cost incurred due to consumer responses and the loss of export revenue due to disease outbreak. There are potentially other losses that would occur along the value chain but to avoid the danger of double counting costs, only potential losses by the final consumer of UK meat and animal products were included. That is the local final consumers and exports that represent external final consumers.

Monetisation of all impacts was then carried out using assumptions as detailed in ANNEX 1a. The final step consists of computing the net present value ( NPV) _³ at the present time using an appropriate discount rate. The flows of costs and benefits associated with disease control measures take place over time. Discountin g future costs and benefits is necessary so that all costs and benefits are expressed in a common metric: the present value.

Presentation of data supporting conclusions

Incursion scenarios results:

The results suggested that the epidemiological and economic analyses should focus on five incursion scenarios:

1. northwards spread, with BTV arriving in April 2009
2. northwards spread, with BTV arriving in July 2008
3. northwards spread, with BTV arriving in September 2008
4. import of infected animals in April 2009
5. import of infected animals in September 2008

The months selected for northwards spread allow for the likely time of arrival in Scotland (autumn 2008 or spring 2009, following spread and over wintering in 2008). The incursion in July 2008 allows for the possibility that BTV could spread more rapidly than expected and, furthermore, coincides with the warmest temperatures and, hence, the greatest potential for spread. The months selected for incursion via imported animals reflect the peaks in the number of movements for both cattle and sheep. Although such an incursion could occur anywhere in Scotland, three counties (Aberdeenshire, Dumfries and Stirling) were particularly at risk because of the high number of movements to these counties.

One potential incursion scenario worth further consideration relates to Northern Ireland. This currently presents no risk, because BTV is absent. If, however, BTV were to spread to Northern Ireland, this would pose a distinct incursion risk to Scotland. All the remaining scenarios were discounted because they posed at most a low level of risk. These include direct introduction of BTV by wind-borne midges from southeast England or continental Europe (see ANNEX 2 (b)), or import of animals at other times of the year. However, this does not mean that these scenarios pose no risk and could not potentially occur.

The accumulated degree-hour EIP model suggested that the seasonal risk of vector-host transmission, a necessary requirement for onward transmission (and, hence, the declaration of a control zone under EU regulations) following introduction, is minimal between December and May-July in Scotland. The duration of the transmission-free period during 2006-2007 varied between 140 days to over 200 depending on region (see ANNEX 2 (c)) .

Epidemiological scenarios results:

Several features are apparent from the results for each incursion/control scenario (see ANNEX 3 (b)). For most scenarios infection seldom spreads after the initial incursion; only if an incursion occurred via northwards spread in July did outbreaks become more widespread in a substantial number of replicates. If it goes undetected, an import of infected animals is likely to result in large-scale outbreak, regardless of the time of the year when the import occurs. For all scenarios without vaccination, the pattern of spatial spread for those replicates which take-off reflects the density of livestock in Scotland.

For all the incursion scenarios, vaccination was efficient at controlling the spread of BTV. Prophylactic vaccination was more effective than reactive vaccination in preventing the initial spread of BTV following an incursion. If the incursion occurred in an area that had a high level (>80%) of vaccine uptake, infection seldom became widespread. Consequently, barrier vaccination with a high level of coverage (>80%) had the greatest impact in the case of incursion via to northwards spread. By contrast, a higher level of spatial coverage at a lower level of uptake was more effective in preventing spread in the case of incursion following an import of infected animals (i.e. where it is difficult to predict the location of the initial incursion). ( Vaccine was assumed to be 100% effective in this model given we had no information to contradict this.)

Vaccination also had a marked impact on the longer-term dynamics of BTV. When only minimal control measures were applied, BTV was still present after two years in a number of replicates for which there was spread following the initial incursion. This was not the case if vaccination was used: infection died out in almost all replicates by the end of the two-year period over which the model was simulated.

Scottish vector data results: (i) Relationship between abundance of farm-associated midge vectors across Scotland and habitat and micro-climatic factors

To date, no relationships could be detected between the abundance of either the C. obsoletus or the C. pulicaris complex and environmental variation measured across Scotland - probably because the 'snapshot' of vector data currently available covers only the tail-end of the season's adult vector activity. However, the numbers of these two complexes were positively correlated with each other across farms - perhaps providing initial indications that these complexes are responding to similar climatic, host and landscape factors (Figure 3 below). The C. obsoletus complex tended to occur in higher abundance than the C. pulicaris complex by around 200 individuals.

Fig. 3. Relationship between the abundance of the C. obsoletus and C. pulicaris groups across sites.

A future approach for predicting when and where in Scotland substantial populations of competent biting midge vectors may occur is outlined for application to the emerging seasonal datasets for farm-associated Culicoides vectors in Scotland (from the parallel RERAD project lead by APS Ltd).

.Culicoides impunctatus(ii) Coincidence of favourable habitat conditions for the Scottish biting midge,

Table 2, below, lists the habitat conditions, in terms of the soil characteristics, landcover and vegetation types preferred by Culicoides impunctatus according to existing literature. This species generally prefers bog or heathland habitat rather than pasture and acidic soils with high organic and water content. When environmental layers corresponding to these characteristics were overlaid, the north-west of Scotland (Cairngorms, northern Grampians and Wester Ross, southern Skye) and Perthshire could be delineated as being at the highest risk of supporting large C. impunctatus populations (having most favourable habitat characteristics) whilst lowland and eastern areas (in Aberdeenshire, the Moray Coast) were at low risk of doing so ( ANNEX 4 (a) i). Farms were situated in medium to high risk areas for C. impunctatus in the north west Highlands, along the Great Glen and at the foot of the Grampians in Aberdeenshire, and in the Borders. These represent areas where 'hand-overs' of BTV may be particularly likely both between farm-associated vectors and C. impunctatus and between domestic ruminants and red deer . These habitat maps need ideally to be interpreted alongside data on the frequency with which C. impunctatus bites large ruminants, can replicate virus to transmissible levels and survive to complete sufficient reproductive cycles for transmission to occur. ( ANNEX 4 (a) ii).

Table 2. Habitat preferences of the Scottish biting midge Culiciodes impunctatus

(iii) ruminants and likely areas of interactions between hostsBTVOverlap of wild and domestics hosts for

When AgriCensus data were mapped, densities of sheep and cattle densitites are highest in lowland areas in Southern and central Scotland, in the north-east, in Angus and Aberdeenshire border areas and in most northern part of Scotland ( ANNEX 4 (b)). Within the main range of red deer, domestic ruminants overlap with large red deer populations on the Cairngorm plateau, along the Moray coast and sporadically through Highland areas ANNEX 4 (c). Despite low average densities of sheep and cattle in the Highlands, the seasonal movement from the lowlands to the Highlands in spring and the reverse in autumn is of potential epidemiological significance - due to the potential for spread of virus to new areas and the difficulty of monitoring infection and disease in livestock that are ranging more widely alongside deer populations.

Economic results:

A worthwhile investment in BTV prevention (baseline surveillance costs C ₃) in each scenario should generate sufficient benefits to at least cover the investment costs. This implies that the net present value ( NPV), the expected net benefit ( ENB) ^₄ should be positive and the benefit-cost ratio ( BCR) greater than one. In economic terms, the higher the values of NPV, ENB and BCR, the more attractive the investment in the baseline surveillance/prevention costs of BTV. Figures 1a to 2b provide the average current discounted benefits (£m) of avoiding BTV incursion for each incursion scenario (a to e see ANNEX 1b) depending on whether the epidemic is constant through years 3 to 5 ( Figure 1a and 1b) or declines ( Figure 2a and 2b) and whether a licence is available to move to slaughter ( Figure 1b) or not ( Figure 2b). Full details including baseline costs, NPV and BCR are tabulated for each incursion scenario in ANNEX 1c. Within incursion scenario, CBA indicators are of the unweighted (without probabilities of BTV incursion R) options generated from the economic spreadsheet model for all scenarios using average epidemiological outcomes only (extreme cases- 5th and 95th percentiles are reported in ANNEX 1d and ANNEX 1e respectively). Since there are considerable uncertainties over assumed probabilities, unweighted benefit cost ratios were used to rank interventions in term of economic efficiency. Results of the weighted scenarios (with estimated probabilities of BTV incursion) are presented in ANNEX 1f.

In the following section, some interpretations of within incursion scenarios (a to e) and the control options C1 to C5 are provided in bullet points.

Figure 1a: Current 5-year Discounted Benefits (£m) of avoiding BTV incursions a - e (constant outbreak no slaughter licence)

Figure 1b: Current 5-year Discounted Benefits (£m) of avoiding BTV incursions a - e (constant outbreak with slaughter licence)

Figure 2a: Current 5-year Discounted Benefits (£m) of avoiding BTV incursions a - e (outbreak dies out no slaughter licence)

Figure 2b: Current 5-year Discounted Benefits (£m) of avoiding BTV incursions a - e (outbreak dies out with slaughter licence)

KEY for Figures 1 & 2: Control Option 1 (C1) = 'Do nothing'; Control Option 2 (C2) = Border Protection Zone ( PZ) with 100% vaccination; Control Option 3 (C3) = PZ to Highland B/F Line with 80% vaccination; Control Option 4 (C4) = PZ whole of Scotland with 50% vaccination; Control Option 5 (C5) = PZ of 100km around incursion with 80% vaccination within PZ

Within Incursion Scenario CBA Analysis

Incursion a (Midge transmission from south in April 2009)

• Total current discounted benefits for five years ranged between £330m - £471m for all incursion scenarios in a. (see ANNEX1c)
• Scenario C2a yields the highest discounted benefit i.e. the greatest return to baseline surveillance costs C ₃. However, as benefits are defined as total disease losses avoided, this control option is associated with the highest disease losses. This means that if this incursion does take place, control option C2 is associated with the highest outbreak costs. (see ANNEX1h).
• The lowest return to baseline surveillance costs C ₃. depends on obtaining a license for movement to slaughter. C4a delivers lowest return with no license to slaughter while C1a delivers lowest return with license to slaughter. These options have the lowest (avoided) disease losses (benefits) and their BCRs are therefore highlighted in bold in ANNEX1c.
• ANNEX1g gives a breakdown of BTV costs using this incursion scenario with control option C4 as an example. This is considered by the team to be the most likely incursion scenario (see ANNEX1a) combined with the best control option (lowest benefits i.e. disease losses avoided). In this case the direct cost for sheep is more than £5M and for cattle more than £20M. Notice that although the direct outbreak costs are higher for sheep than cattle, these are dwarfed by the other direct costs, which are dominated by direct costs for cattle. These direct costs for cattle are mainly the unavoidable baseline veterinary and medicine costs (see ANNEX1a). Notice too that indirect costs (lost markets etc.) far exceed direct costs.

Incursion b (Midge transmission from south in July 2008)

• Total current discounted benefit for five years ranged between £338m - £468m. (See ANNEX1c)
• As in incursion scenario a, scenario C2b yields the greatest returns on investment in baseline surveillance costs C ₃ (highest outbreak losses avoided).
• Unlike incursion scenario a, the lowest return to C ₃ (lowest outbreak losses avoided) depends on the duration of the outbreak or the trajectory by which the disease persists after an outbreak as well as on the license position. Scenario C4b gives lowest disease losses avoided (best vaccination strategy) when BTV lingers and causes losses equivalent to year 2 levels up to year 5. C1b does the same when BTV gradually dies out after year 2 with no licence to slaughter. However, C4b remains the best vaccination strategy as C1b is the no vaccination option.

Incursion c (Midge transmission from south in September 2008)

• Total current discounted benefits for five years ranged between £336m - £454m. (See ANNEX1c)
• As in incursion scenarios a and b, scenario C2c yields the most returns to C ₃ (highest outbreak losses avoided).
• As in scenario a, the lowest return to C ₃ (lowest disease losses, best disease control) depends on obtaining a license for movement to slaughter. C4c gives lowest return to C ₃ with no license to slaughter while C1c gives lowest return with license to slaughter.

Incursion d (Animal import April 2009)

• Total current discounted benefits for five years ranged between £339m - £500m.(See ANNEX1c)
• As in scenarios a, b and c, scenario C2d yields the most returns to C ₃ (highest outbreak losses avoided).
• The lowest return to C ₃ (lowest outbreak losses avoided) in scenario d appeared not to be influenced by either license for move to slaughter or the time trajectory of the disease when an incursion occurs. The lowest returns on baseline costs C ₃ remained scenario C1d for all treatment options. However, as this was the 'no vaccination' strategy, C4d remained the vaccination strategy with the lowest disease losses avoided.

Incursion e (Animal import September 2008)

• Total current discounted benefits for five years ranged between £334m - £478m. (See ANNEX1c)
• As in scenario a, b, c and d, scenario C2e yields the most returns on investment in baseline surveillance and control costs C ₃.
• As in scenario d, the lowest return to C ₃ and therefore the lowest avoided disease losses appeared not to be influenced by license for move-to-slaughter or the time trajectory of the disease when an incursion occurs. The lowest returns on investment costs C ₃ remained scenario C1e for incursion scenarios. The vaccination strategy with the lowest returns on investment costs C ₃ and therefore the lowest disease losses was C5e.

Summary of 'within' incursion Scenario Analysis

As the relative risk of alternative incursion scenarios is uncertain, it was useful to assess alternative control options within scenarios. The higher a CBA ratio within a scenario, the more favourable is that investment in economic efficiency terms. In all scenarios, the ranking of CB ratios and sum of current discounted benefits indicates that control option C2 (see Figures in the ANNEX1c) yields the most returns on investment. In other words, if the £141m discounted costs of disease prevention are successful and BTV is avoided then C2 would give the greatest discounted current benefits (costs avoided). This means that C2 is associated with the greatest total disease losses. Maintaining the same assumptions but looking at the extreme cases- 5 ^th and 95 ^th percentiles of the CBA (see ANNEX1d and 1e for results) shows that control option C2 again is associated with the highest disease outbreak losses (greatest benefit if avoided and therefore greatest return to C ₃) while C1 is associated with the lowest disease outbreak losses (lowest benefit if avoided and therefore lowest return to C ₃l). As C1 is the 'do nothing' option the 'cure' is often more expensive than the disease i.e. output losses under C1 must often be less than the total disease losses (lowered output losses plus control costs) under other control options. However, as C4 ( PZ all Scotland, 50% vaccination) often provides the lowest avoided costs (benefits) or runs a close second to C1, this is the best vaccination strategy examined. The only exception to this rule seems to be incursion e (animal import Sept 08) where the localised vaccination strategy (C5e) gives the lowest avoided total disease losses.

Sensitivity Analysis

The sensitivity analysis addresses the presence of uncertainty in the CBA based on the key assumed parameters adopted. In essence, sensitivity analysis proposes "what if" scenarios by manipulating certain variables to determine minimum and maximum values of the analytic measures. In this way, the CBA becomes more robust concerning any challenges to its original assumptions. Sensitivity analysis was conducted for the most uncertain parameters used in the CBA analysis. Sensitivity analysis was conducted for changes of ±2% and ±5% for the following parameters: weight loss; milk loss; fertility loss; wool loss; export multipliers for cattle and sheep; own-price elasticities for sheepmeat, beef and milk.

The 'means' & 'q95' & 'q5' CBA:

The sensitivity analysis indicated no change in the CBA ratios for all the three cases (means, q95, q5) when changes of ±2% and ±5% were applied for weight loss, milk loss, fertility loss and wool loss. Again, no change in the CBA ratios when changes of ±2% were applied for export multipliers for cattle and sheep and own-price elasticities for sheepmeat, beef and milk. However, the sensitivity analysis showed a change of ±1% in the CBA ratios when export multipliers for cattle and sheep and own-price elasticities for sheepmeat, beef and milk were varied by ±5%.

4.2 Links/Institutional co-ordination (for projects at several centres).

The project was led by Prof. George Gunn from SAC who is one of the EPICPIs for EPIC Module 1. This Module includes the subject area covered by the objectives of this current study. George co-ordinated the work of four overlapping teams within the project. These teams were:

Epidemiology Modelling led by Dr Simon Gubbins, IAH Pirbright

Animal Health Economics led by Dr Alistair Stott, SAC

Vector Information led by Dr Beth Purse, CEH

Expert Panel led by Dr Kathy Johnston, SG

The Epidemiology team was primarily IAH based but included staff from the Met. Office.

The Economics team were SAC staff but they liaised closely with IAH staff and both teams worked with BioSS to agree the epidemiology output interface with economics. The Vector team included close collaboration with Advanced Pest Solutions and Macaulay Institute. The vector team liaised closely with the Epidemiology team. The full project was managed through a series of three expert panel meetings and these were organised by Kathy Johnston in liaison with George Gunn. The project staff listed above attended each meeting but was joined by colleagues from Edinburgh University, Moredun Research Institute and Glasgow University in addition to advisors from Scottish Government and ultimately representatives of NFU Scotland and the Scottish Livestock Markets.

4.3 Summary of results

Incursion scenarios outputs

• The most likely incursion scenarios are northwards spread from south-east England or import of infected animals.
• The risk of direct incursion of infected vectors from affected areas in south-east England or mainland Europe is very low, but not negligible.
• If a focus of infection were to become established in the north of England or Northern Ireland, this would pose a distinct incursion risk for Scotland.

Epidemiological outputs

• Under most scenarios infection seldom spread after the initial incursion; only if the incursion occurred in July did outbreaks become more widespread in a substantial number of replicates.
• Barrier vaccination at a high level of uptake had the greatest impact on the incursions via northwards spread.
• However, a higher level of spatial coverage with a lower level of uptake was most effective at controlling an incursion via imported animals.

Scottish vector data outputs

• Maps of suitable habitat for the bog-heathland Scottish biting midge - Culicoides impunctatus have been produced. The northern uplands are at high risk of supporting large C. impunctatus populations.
• The Scottish biting midge is most likely to overlap with farmland, and with domestic ruminants and farm associated vectors north west Highlands, along the Great Glen and at the foot of the Grampians, and in the Borders
• Domestic ruminants overlap with large red deer populations on the Cairngorm plateau, along the Moray coast and sporadically through Highland areas.
• We could not predict how the numbers of farm-associated midge vectors ( C. obsoletus or the C. pulicaris complexes ) - is likely to vary across Scotland on the basis of current vector surveillance data. We outline a future framework for such predictions.

Economics Outputs

• Economic analysis is based primarily on the average discounted benefits of avoiding BTV8 in Scotland due to baseline surveillance costs.
• It was not possible to estimate the probability of incursion under each scenario modelled because the epidemiologists did not believe sufficient information was available to make such estimates.
• Higher economic indicators show best return to baseline surveillance costs. This was always control option C2 (border PZ, 100% vaccinated) i.e. C2 was associated with highest disease incursion losses avoided. It follows that this vaccination strategy is associated with high total disease losses.
• Assuming the incursion investigated does take place, either vaccination of all Scotland at 50% uptake (C4) or 'no vaccination' (C1) offer the lowest total disease losses (average discounted benefits of disease losses avoided).
• At the extremes of the epidemiological output (95 ^th and 5 ^th percentiles) the lowest average discounted benefits (disease losses avoided) were with control option C1 regardless of incursion scenario, with option C4 usually giving the next lowest disease losses avoided.
• Sensitivity analysis showed no major impacts on the cost benefit analysis results after ±5% change in key assumptions.

« Previous | Contents | Next »