An Economic Evaluation of Preclinical Testing Strategies Compared to the Compulsory Scrapie Flock Scheme in the Control of Classical Scrapie

Abstract

Cost-benefit is rarely combined with nonlinear dynamic models when evaluating control options for infectious diseases. The current strategy for scrapie in Great Britain requires that all genetically susceptible livestock in affected flocks be culled (Compulsory Scrapie Flock Scheme or CSFS). However, this results in the removal of many healthy sheep, and a recently developed pre-clinical test for scrapie now offers a strategy based on disease detection. We explore the flock level cost-effectiveness of scrapie control using a deterministic transmission model and industry estimates of costs associated with genotype testing, pre-clinical tests and the value of a sheep culled. Benefit was measured in terms of the reduction in the number of infected sheep sold on, compared to a baseline strategy of doing nothing, using Incremental Cost Effectiveness analysis to compare across strategies. As market data was not available for pre-clinical testing, a threshold analysis was used to set a unit-cost giving equal costs for CSFS and multiple pre-clinical testing (MT, one test each year for three consecutive years). Assuming a 40% within-flock proportion of susceptible genotypes and a test sensitivity of 90%, a single test (ST) was cheaper but less effective than either the CSFS or MT strategies (30 infected-sales-averted over the lifetime of the average epidemic). The MT strategy was slightly less effective than the CSFS and would be a dominated strategy unless preclinical testing was cheaper than the threshold price of £6.28, but may be appropriate for flocks with particularly valuable livestock. Though the ST is not currently recommended, the proportion of susceptible genotypes in the national flock is likely to continue to decrease; this may eventually make it a cost-effective alternative to the MT or CSFS.

Full text

An Economic Evaluation of Preclinical Testing Strategies
Compared to the Compulsory Scrapie Flock Scheme in
the Control of Classical Scrapie
Lisa Boden
1.
, Ian Handel
2.
, Neil Hawkins
3
, Fiona Houston
1
, Helen Fryer
4
, Rowland Kao
1
*
1Institute of Biodiversity, Animal Health and Comparative Medicine, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, United Kingdom,
2The Roslin Institute and Royal (Dick) School of Veterinary Studies, University of Edinburgh, Midlothian, United Kingdom, 3Oxford Outcomes, Oxford, United Kingdom,
4Department of Zoology, The Institute for Emerging Infections, The Oxford Martin School, Oxford University, United Kingdom
Abstract
Cost-benefit is rarely combined with nonlinear dynamic models when evaluating control options for infectious diseases. The
current strategy for scrapie in Great Britain requires that all genetically susceptible livestock in affected flocks be culled
(Compulsory Scrapie Flock Scheme or CSFS). However, this results in the removal of many healthy sheep, and a recently
developed pre-clinical test for scrapie now offers a strategy based on disease detection. We explore the flock level cost-
effectiveness of scrapie control using a deterministic transmission model and industry estimates of costs associated with
genotype testing, pre-clinical tests and the value of a sheep culled. Benefit was measured in terms of the reduction in the
number of infected sheep sold on, compared to a baseline strategy of doing nothing, using Incremental Cost Effectiveness
analysis to compare across strategies. As market data was not available for pre-clinical testing, a threshold analysis was used
to set a unit-cost giving equal costs for CSFS and multiple pre-clinical testing (MT, one test each year for three consecutive
years). Assuming a 40% within-flock proportion of susceptible genotypes and a test sensitivity of 90%, a single test (ST) was
cheaper but less effective than either the CSFS or MT strategies (30 infected-sales-averted over the lifetime of the average
epidemic). The MT strategy was slightly less effective than the CSFS and would be a dominated strategy unless preclinical
testing was cheaper than the threshold price of £6.28, but may be appropriate for flocks with particularly valuable livestock.
Though the ST is not currently recommended, the proportion of susceptible genotypes in the national flock is likely to
continue to decrease; this may eventually make it a cost-effective alternative to the MT or CSFS.
Citation: Boden L, Handel I, Hawkins N, Houston F, Fryer H, et al. (2012) An Economic Evaluation of Preclinical Testing Strategies Compared to the Compulsory
Scrapie Flock Scheme in the Control of Classical Scrapie. PLoS ONE 7(3): e32884. doi:10.1371/journal.pone.0032884
Editor: Michael George Roberts, Massey University, New Zealand
Received June 23, 2011; Accepted February 6, 2012; Published March 7, 2012
Copyright: ß2012 Boden et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: LB is funded by Defra project (Department of Farming and Rural Affairs) SE0250. IH is funded by BBSRC (Biotechnology and Biological Sciences
Research Council) institutional strategic programme grant 338BDD RA0762. RRK is funded by a Wellcome Trust Senior Research Fellowship. The funders had no
role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: rowland.kao@glasgow.ac.uk
.These authors contributed equally to this work as first authors.
Introduction
Economic evaluations are a well-accepted component of the
evaluation of policies to manage chronic diseases in human
populations [1]. Without them, it is difficult to make a useful
contribution to decision-making on disease control policy [1].
Despite this, they are often not undertaken due to the complexity
and multitude of consequences (such as animal welfare, environ-
mental protection and food security [1]) for which there often are
no readily available market estimates of financial value, especially
when combined with the complexities inherent in infectious
disease dynamics. Additionally, as economic evaluations in
healthcare are most easily interpreted when based upon
experimental studies that evaluate the effectiveness of alternative
strategies [2], there may be some reluctance to accept them when
they use data generated from epidemiological models where
nonlinear dynamics introduce additional uncertainties into the
decision-making process. However, policy or individual consump-
tion decisions (including decisions to invest resources in collecting
more evidence) still need to be made given currently available
data, and economic evaluations can help make this decision-
making more efficient (for example by highlighting opportunity
costs).
In veterinary epidemiological research many economic evalu-
ations are based on comparisons using cost analysis (only
examining costs of options) or cost minimisation (examining costs
of options assuming equivalent benefits) techniques [2]. Some
veterinary studies utilise true cost-effectiveness (measuring both
costs and benefits) or cost-benefit analyses (measure of costs and
non-equivalent effects where benefits are measured in monetary
units) [3–6]. Few veterinary studies incorporate cost-utility
analyses (benefits are measured QALYs or DALYs or other utility
scale) such as found in the medical literature [7]. This study is a
cost-effectiveness analysis which examines the costs and the
benefits of competing scrapie control strategies considered from
a societal perspective.
One example where these issues are particularly pertinent is the
control of scrapie in sheep and goats. Scrapie is a transmissible
spongiform encephalopathy (TSE) which results in an invariably
fatal, progressive neurodegenerative disease of sheep, goats and
PLoS ONE | www.plosone.org 1 March 2012 | Volume 7 | Issue 3 | e32884

moufflon. It is associated with an abnormal form of the prion
protein (PrPSc) [8]. Other distinct transmissible spongiform
encephalopathies (TSEs) have been recognized as occurring
separately in humans and animals including bovine spongiform
encephalopathy (BSE) (first recognized in 1986) and a new variant
of Creutzfeldt-Jakob Disease (CJD) (1996). A possible link between
bovine spongiform encephalopathy (BSE) in cattle and variant
Creutzfeldt-Jakob Disease (vCJD) in humans [9–12] has resulted
in an increased prioritisation of scrapie eradication in the EU and
thus Great Britain (GB).
In 2001, this was acted upon in Great Britain (GB) via the
National Scrapie Plan (NSP) [13,14].The NSP’s primary objectives
were to eradicate scrapie and breed for TSE resistance in the
national sheep flock [15], thereby minimizing the likelihood that
BSE could be present and not detected in the national flock and
diminishing the incidence of scrapie in the process [16]. At the time
of the NSP’s inception, there were no cheap or effective pre-clinical
diagnostic tests available and there was speculation that the possibly
low incidence of BSE in sheep might have been masked by the
presence of scrapie. As a result, a genetically-based breeding
strategy targeting susceptibility, rather than disease, was thought to
provide the most reasonable chance for success [16]. In 2004, the
NSP was augmented by a slaughter and replacement scheme.
Initially, this was a voluntary programme, but after July 2004,
control became mandatory for all flocks with confirmed cases from
that date (Compulsory Scrapie Flock Scheme (CSFS)), as required
by EC Regulations [17–19]. Although the CSFS is undoubtedly
effective in scrapie eradication, it is not applied uniformly to all
flocks within GB, as testing the national flock would be prohibitively
expensive and large numbers of healthy sheep would be culled.
The development of a live test for scrapie [20] suggests that pre-
clinical testing may now provide a more cost-effective disease-
based strategy for scrapie eradication in the UK. The efficacy of
adopting a strategy aimed at controlling disease by targeting
infected sheep rather than targeting sheep at risk of being infected
(due to the susceptibility of their genotype or being from an
affected flock) was explored by Boden and colleagues [16]. In that
study, a deterministic within-flock model was used to demonstrate
that only large flocks with a large proportion of homebred
breeding sheep are likely to be a significant risk for onward flock-
to-flock transmission of scrapie. For most flocks it was found that
the CSFS could be replaced by a strategy using a currently
available live test without excessive risk to other farmer’s stock,
even if the proportion of susceptible genotypes in the flock is
unusually large. Even for flocks that represent a high risk of
harbouring a high prevalence of infection, there would be limited
probability of onward transmission if scrapie is detected soon after
disease introduction (typically less than 5 years). However, if
detection of disease is delayed, onward transmission remains a
concern and it may be more appropriate to retain the existing
CSFS strategy in these flocks.
In this study we compare the direct costs and the effects of
different pre-clinical testing strategies for classical scrapie.
Although previous studies have considered scrapie control at the
flock [11,21–24] and national flock level [14,25,26], few have
examined the economics of scrapie control policies [27,28] and no
studies have explicitly looked at the cost effectiveness of
implementing such policies.
Materials and Methods
Model of within-flock scrapie transmission
Within-flock scrapie prevalence after implementation of the CSFS and pre-
clinical diagnostic testing strategies
A difference equation model was used to describe the within-
flock spread of scrapie for three different classifications of sheep
flocks within the UK [16,23]:
NHigh risk flocks (large purebred and commercial flocks ($500
sheep) with large proportions of homebred sheep ($0.89)
NMedium risk flocks (large commercial flocks ($500 sheep) with
small proportions of homebred sheep (#0.10)).
NLow risk flocks (small purebred and commercial flocks (#200
sheep) with large ($0.89) and small proportions (#0.10) of
homebred sheep.
Lambing management and infected placental material are
believed to play a large part in the transmission of scrapie,
therefore flock classifications were based on the risk that scrapie
affected flocks posed to other flocks through the onward sale of
breeding sheep [16]. A full description of the model variables,
parameters and equations is presented in Fryer et al. 2007 [23],
reproduced in Boden et al. 2010 [16].
The model was adapted to consider the impact of three
proposed control and eradication strategies for scrapie and
contrasted with a strategy where no intervention occurs for up
to 15 years, (the estimated average duration of a within-flock
epidemic) [29].
There were four strategies compared:
NNo intervention.
NThe current Compulsory Scrapie Flock Scheme (CSFS)
NMultiple pre-clinical diagnostic tests (MT) - once a year for
three years
NSingle pre-clinical diagnostic test (ST) - once at a single time
within a single year.
After implementation of the CSFS and the pre-clinical
diagnostic testing strategies, two years of restrictions are imposed
and genetically resistant replacement sheep were bought in. These
strategies are described in further detail in Boden et al. (2010) [16].
For each flock type, the average prevalence of scrapie per year
was calculated, for every year past the initial intervention. The
average prevalence was converted into number of infected sheep
sold on to other farms by multiplying the prevalence and the
number of infected sheep sold by each flock type. The prevalence
and number of infected sheep sold in each year were summarized
into a single point estimate for each flock by assuming a
probability of detection in each year of a 15 year epidemic
(Figure 1). This assumed that scrapie detection increased 3 years
after initial infection and peaked between 5 and 7 years after initial
infection. Lower probabilities of detection were assigned as the
number of years since infection increased as it was assumed that
the majority of cases of clinical scrapie would be detected between
5–7 years [30]. For each flock type, a distribution of flock size was
obtained from the 2002 scrapie postal survey [31]. This was used
to calculate testing and compensation costs for each of the control
strategies in each of the flock types.
Economic analysis
Cost effectiveness analysis. In the cost-effectiveness
analysis, costs and the effects of each strategy were compared for
each flock type (high, moderate and low risk flocks). The effects of
each strategy were based on the reduction in the onward sale of
infected potential breeding sheep relative to the number sold with
no testing strategy.
The summary measure of cost-effectiveness was the incremental
cost per infected sheep sale avoided. The use of this measure
Economic Evaluation of Scrapie Testing Strategies
PLoS ONE | www.plosone.org 2 March 2012 | Volume 7 | Issue 3 | e32884

implies that societal benefit is a linear function of the number of
infected sheep sold. As only a small proportion of onwards sales
will be to breeding stock we have assumed that the onwards risk of
transmission and hence societal impact is linearly related to the
number of infected sheep sold.
Alternative control strategies were considered according to their
effectiveness and their costs. Any strategies where there was an
alternative that was both cheaper and more effective were
removed from the comparison as these are ‘dominated’ strategies
[2]. Strategies subject to extended dominance [2], where an
alternate strategy can be replace by a more cost effective mix of
alternatives, were not removed as we made the assumption that
strategies may have to be applied to either all flocks nationally, or
all flocks in a risk stratum.
Costs
The sum of the total costs for travel, time (veterinarian and
helper), sampling consumables, testing, examination, and report-
ing for genotype testing and costs of compensating the farmer for
sheep removed from the flock were calculated for each strategy for
each year since the flock was infected (in a 15 year epidemic). A
description of these costs is outlined in Table 1. As market-based
estimates of the cost of the pre-clinical test were not available, a
threshold analysis was performed (using the base scenario with a
test sensitivity of 90% and prevalence of 40% susceptible
genotypes in the high risk flocks). The pre-clinical test was initially
assumed to have the same unit cost and volume discounting costs
of the genotyping. This unit cost was then multiplicatively scaled to
identify the threshold at which the total cost of the multiple testing
strategy was equal to the total cost of the CSFS.
The ‘average’ unit cost of pre-clinical testing at the threshold
price was calculated by dividing the sum of the cost of testing all
flocks by total number of sheep tested. Cost Effectiveness analysis
was subsequently used to compare the strategies where applicable.
Sensitivity of the results to preclinical test cost was explored by
repeating the analysis with the preclinical test unit cost set at 0.5
and 1.5 times the threshold value.
Assumptions
We assumed that the cost and effort of replacing stock (culled as
a consequence of removal from the flock due to scrapie) in the two
years following each of the three strategies was equivalent to the
current level of compensation offered by Defra. The costs of
scrapie testing (genotype or pre-clinical testing) and compensation
were assumed to be constant over each year of an epidemic (i.e.
cost of compensation did not depend on prevalence of scrapie in
flock). We assumed that the proportion of susceptible or diseased
sheep in each flock would be distributed equally across the
proportions of ewes and rams in eafch flock.
The additional value of pedigree and pregnant sheep were not
factored into these analyses. The calculation of the costs of the
testing strategies did not take into account the costs for shipment of
samples from base to the laboratory where samples were
examined. Additionally these costs did not take into account the
costs for the time of the farmer/stockmen to help handling the
samples.
Consequences
Reduction in the number of infected sheep sold. The
number of infected sheep sold by each flock was calculated from
the deterministic model [16,23] parameterised with data from the
scrapie postal survey [31]. The number of infected sheep sold for
each control strategy was then subtracted from the number of
infected sheep sold would there to be no control strategy to give
the reduction in the number of infected sheep sold to other farms.
The number of infected sheep sold after implementation of the
CSFS is always assumed to be zero as this strategy removes all
susceptible genotypes within the flock.
Sensitivity analyses
The base analysis assumed a pre-clinical test sensitivity of 90%
and that 40% of each flock had a scrapie susceptible genotype.
The within-flock model was re-run and economic analyses were
repeated to allow for differences in the test sensitivity (70%, 50%)
and proportions of susceptible genotypes in the flock (30%,
20%,10%).
The model was initially implemented in excel and then
independently written in R statistical software to validate the
results.
Results
The effectiveness of a strategy, as measured by avoiding the
onwards sale of infected sheep compared to no control strategy, is
a function of the strategy, the flock risk class and the proportion of
susceptible sheep within the flock (and test sensitivity for the pre-
clinical testing based strategies). Based on the base costings, the
contribution of compensation costs for culled animals was 57% for
CSFS, 19% for MT and 10% for ST. Cost and effectiveness results
at the threshold cost for preclinical testing are shown in Figure 2.
For each sensitivity analysis, scenario ST was the cheapest strategy
at the threshold unit cost for preclinical testing. Likewise ST was
least effective and CSFS most effective. For all but the high risk
flocks only low (,0.7) numbers of infected sheep sales were
avoided. The resulting incremental costs effectiveness ratios in the
low and medium risk flocks were relatively high ranging from
£12,000 per sale avoided to £2.6 M per sale avoided, with the
upper end representing the high cost of attempting control in
effectively resistant flocks (R
0
below one). Unless the value of
avoiding sales of infected sheep exceeds £12,000 none of the three
testing strategies would be considered cost-effective in the low and
medium risk flocks. The following description and discussion of the
results will focus on the high-risk flocks.
To achieve equivalent overall costs for MT and CSFS in the
high-risk flocks, assuming within-flock prevalence of 40%
susceptible genotypes and a test sensitivity of 90% the ‘average’
unit cost of pre-clinical testing had to be set at £6.28.
Figure 1. The assumed distribution of probability of detection
in each year of a 15 year scrapie epidemic [30].
doi:10.1371/journal.pone.0032884.g001
Economic Evaluation of Scrapie Testing Strategies
PLoS ONE | www.plosone.org 3 March 2012 | Volume 7 | Issue 3 | e32884

Costs, effects and incremental cost effectiveness ratios (ICER –
the incremental cost of avoiding the sale of an additional infected
sheep) are presented in Table 2 for the base scenario with a test
sensitivity of 90% when the proportion of susceptible genotypes in
the flock is 40%. The cost/effectiveness of each testing strategy is
shown in Figure 3 for the threshold cost of pre-clinical testing and
scenarios where pre-clinical testing is 0.5 and 1.5 times its
threshold cost.
In the high-risk flocks, assuming within-flock prevalence of 40%
susceptible genotypes and a test sensitivity of 90%, the ST was the
cheapest (£18,870) but least effective of the three testing strategies (30
infected-sales-averted). The MT (42 infected-sales-averted) was slightly
less effective than the CSFS (44, i.e. all, infected-sales-averted). MT/
CSFS costs for testing and compensation were £50,891.
Compared to no-intervention the ST strategy reduces infected
sheep sales at £629 per sale averted (14.25 infected sheep sold
compared to 44.26). CSFS reduces infected sheep sales to zero at
an incremental cost per sale averted (ICER) of £2247.
If the average unit-cost per sheep of pre-clinical testing was
scaled upwards the MT strategy was dominated, with the CSFS
strategy costing less and reducing infected sheep sales to zero. With
a preclinical test unit cost of 1.5 times the threshold cost (i.e.
£9.41) the ST strategy reduced sales of infected sheep at £741 per
sale averted compared to no intervention and CSFS had an ICER
compared to ST of £741.
If the average unit-cost per sheep of pre-clinical testing was
scaled downwards (0.5 times the equivalent unit-cost of preclinical
testing) no strategies were dominated with ST having a cost per
infected sale avoided of £517, MT compared to ST of £2142 and
CSFS compared to MT of £4136.
In high risk flocks, decreasing the sensitivity of the diagnostic
test reduced the efficacy of the multiple (MT) and single (ST) test
strategies. At lower proportions of susceptible genotypes (i.e.
#30%), MT is the dominated strategy. Thus as test sensitivity and
proportions of susceptible genotypes decreases, the cost of MT has
to be increasingly less than the threshold cost for it to be
considered a reasonable alternative to the CSFS. If the prevalence
of susceptible genotypes decreases over time, the difference in the
effectiveness between the ST and MT/CSFS will become smaller.
Under those conditions, in high risk flocks, the ST could be an
alternative to the MT or CSFS.
Discussion
Thisstudycomparesthecostsandeffectsofadoptinga
diagnostic testing strategy (either single or multiple testing)
instead of the current CSFS to diagnose and eradicate classical
scrapieinGBsheepflocks.Theseanalysesareonlyapplicable
for flocks that have been identified as having at least one clinical
case of scrapie, which would result in the mandatory application
Table 1. Description of costs incurred for Compulsory Scrapie Flock Scheme, single and multiple pre-clinical testing strategies.
Costs Description
Compensation for the compulsory
scrapie flock scheme (CSFS) and
diagnostic testing strategies
The mean number of sheep that the farmer is compensated for is equal to the proportion of susceptible sheep * flock
size.
Where:
a.Proportion of genetically susceptible sheep = 0.40
b.Proportion of ewes in flock = 0.98
c.Proportion of rams in flock = 0.02
d.Cost per ewe £65
e.Cost per ram £90
This cost is constant over each year of an epidemic (i.e. cost of compensation does not depend on prevalence of
scrapie in flock). We assumed that the proportion of susceptible sheep in each flock will be reflected equally across
the proportions of ewes and rams.
Compensation for multiple
and single testing strategies
Compensation is paid for each sheep positive for scrapie as detected by the pre-clinical test. The test is assumed to
have 90% sensitivity and 100% specificity. Therefore, the mean number of sheep detected as positive for scrapie is
dependent on the prevalence of scrapie in the flock in each year since the flock was infected multiplied by 0.90. In the
multiple testing strategy, the entire flock is tested each year for three years and the farmer is compensated for the
total mean number of sheep detected as positive in years 1, 2 and 3. This total number is never greater than the total
number of infected sheep in the year since infection because it is assumed that the prevalence does not substantially
increase in the second and third year of testing (e.g. mean number of sheep detected in year 2 = 0.9* (number of
infected sheep in year 12number of infected sheep detected and removed by the test in year 1).
Proportion of ewes positive for scrapie = 0.98*prevalence
Proportion of rams positive for scrapie = 0.02 *prevalence
Costs per ewe £65
Cost per ram £90
Travel (All strategies From base to farm; 40 pence/mile @ 45 miles/h. We assumed that the average farm was half an hour away from base.
Time (All strategies) Veterinarian = £70/hour; helper = £40/hour; @ 40 samples/hour
Sampling consumables
(All strategies)
Speculum, forceps, scissors, containers = £5/sample
Genotype testing
(CSFS only)
*1–10 samples @ £ 29.00 per sample;
*11–29 samples @ £22.50 per sample;
*Subsequent numbers of sheep test @ £14.50 per sample
Farmer can do the sampling themselves if sheep are not for export.
If sheep are for export, then additional veterinarian/helper charges are also considered in this cost.
Examination of preclinical
test samples
(Testing strategies only)
Market based price data not available. The pre-clinical test was initially assumed to have the same unit cost and
volume discounting costs of the genotyping. This cost was then multiplicatively scaled to identify the threshold at
which the total cost of the MT strategy was equivalent to the total cost of the CSFS strategy in the 90% sensitivity and
40% prevalence scenario. (i.e. the point at which the MT strategy becomes dominated by CSFS).
doi:10.1371/journal.pone.0032884.t001
Economic Evaluation of Scrapie Testing Strategies
PLoS ONE | www.plosone.org 4 March 2012 | Volume 7 | Issue 3 | e32884

of a scrapie control measure (CSFS or one of the testing
strategies).
The efficacy of each of the alternate strategies compared to the
CSFS was examined in Boden et al. (2010) [16] and provided the
data for the efficacy measures used in this study. In that study,
Boden et al. (2010) [16] recommended that for most flocks
(moderate and low risk) the CSFS could be replaced by a strategy
using a currently available live test without excessive risk to other
farmers, even if the proportion of susceptible genotypes in the flock
is unusually large. For high risk flocks, it may be more appropriate
to retain the existing CSFS strategy in these flocks [16].
In this study, it is assumed that with the CSFS, zero infected
animals will be sold. However, based on genotype, some sheep are
more resistant than others and it is possible that a resistant animal
can be infected with scrapie, especially if it is not ARR/ARR. As
such, it is possible that scrapie in non-ARR/ARR sheep may be
retained in the CSFS strategy and some onward movement of
infection between flocks may occur. However, it is expected that in
the current environment, the within-flock R0 for most flocks will
remain low. Additionally, the long incubation periods associated
with more resistant sheep suggest that their contribution to new
outbreaks is small and unlikely to have a major impact on the
results of this study.
This study does not address any costs associated with atypical
scrapie. Even though breeding for resistance to classical scrapie
does not select for resistance to atypical scrapie, selection of ARR
homozygote genotypes is unlikely to increase the prevalence of
atypical scrapie [32] and to date, there has been no evidence that
Figure 2. Cost-effectiveness plane examining all flocks (high, moderate and low risk) with different proportions of susceptible
genotypes. In this analysis, we have found the unit cost of pre-clinical testing that makes the total CSFS and MT (including compensation, labour
and testing) equivalent costs. The cost effectiveness planes vary by test sensitivity (rows 50–90%) and proportions of susceptible genotype (columns
10–40%) The outlined points on the plot represent dominated strategies. The origin in each panel represents no intervention. In high-risk flocks, the
CSFS and MT are more efficient than the single test strategy (ST) at high proportions of susceptible genotypes within the flock (i.e. .30%). At lower
proportions of susceptible genotypes (i.e. #30%) in high risk flocks, MT is the dominated strategy. MT has to be cheaper, and this is exacerbated if
prevalence or test sensitivity drops.
doi:10.1371/journal.pone.0032884.g002
Economic Evaluation of Scrapie Testing Strategies
PLoS ONE | www.plosone.org 5 March 2012 | Volume 7 | Issue 3 | e32884

Table 2. Costs and effects (infected-sales-averted) for each scrapie control strategy assuming Defra pay all costs.
Preclin test cost Strategy
Mean Infected
Sales Per Flock
Incrmntal
Infected
Sales Avoided
Mean Number
of Sheep Tested
Mean number of
sheep slaughtered
Cost of Sheep
Slaughtered
Cost of
Testing
Total
Cost
Incrmntal
Cost (GBP)
Incrmntal
Effect
ICER (cost per
infected sale
avoided)
base None 44.26 0 0 0 0 0 0 -
(£6.28 unit cost) ST 14.25 30.01 1109 54 3535 15335 18870 18870 30.01 629
MT 2.44 41.82 3327 75 4885 46006 50891 32021 11.81 dominated
CSFS 0 44.26 1109 444 29068 21823 50891 32021 14.25 2247
preclin testing 61.5 None 44.26 0 0 0 0 0 0 -
(£9.42 unit cost) ST 14.25 30.01 1109 54 3535 18698 22233 22233 30.01 741
MT 2.44 41.82 3327 75 4885 56098 60983 38750 11.81 dominated
CSFS 0 44.26 1109 444 29068 21823 50891 28658 14.25 2011
preclin testing 60.5 None 44.26 0 0 0 0 0 0 -
(£3.14 unit cost) ST 14.25 30.01 1109 54 3535 11971 15506 15506 30.01 517
MT 2.44 41.82 3327 75 4885 35914 40799 25293 11.81 2142
CSFS 0 44.26 1109 444 29068 21823 50891 10092 2.44 4136
This illustrates the costs and effects for high risk farms using CSFS and pre-clinical tests with a sensitivity of 90%, when the proportion of genotypes in the flock is 40%. The incremental cost effectiveness ratio (ICER) is also
presented where appropriate.
doi:10.1371/journal.pone.0032884.t002
Economic Evaluation of Scrapie Testing Strategies
PLoS ONE | www.plosone.org 6 March 2012 | Volume 7 | Issue 3 | e32884

atypical scrapie is either transmissible or a potential zoonosis [33–
35]. Therefore selection towards resistance to classical scrapie is
unlikely to have meaningful economic consequences, even if it
does result in a small increase atypical scrapie prevalence.
The results of this economic analysis support the replacement of
the CSFS with a single testing strategy in low and moderate risk
flocks if a control strategy is mandatory and the testing and
compensation costs are borne by Defra; in these instances, a single
pre-clinical diagnostic test strategy is the cheapest strategy
compared to the MT and the CSFS.
In high risk flocks, at high proportions of susceptible genotypes
and high test sensitivity, CSFS strategies would be recommended if
the additional benefits of avoiding the onwards sale of approxi-
mately 14 sheep obtained by CSFS were worth at least the
estimated £2247 per sale averted.
The CSFS strategy results in zero onwards infected sheep sales
(2.44 less than the MT strategy) however if the unit cost of
preclinical testing were sufficiently low the additional cost of
avoiding these sales may be judged unacceptable and MT may be
the preferred strategy. For example if the preclinical test unit cost
was £3.14 per sheep, on average, the reduction in onwards
infected sheep sales from 2.44 to zero would cost £4136 per sale
averted. A move from CSFS raises the issue of disinvestment. In
economic evaluation this is an apparent asymmetry of decision
making whereby interventions are adopted at a lower cost per unit
of benefit than they are abandoned. It may be that moving to a
strategy which is only slightly less effective (MT) may be
considered unacceptable even it markedly reduces costs.
If the preclinical testing unit cost was greater than or equal to
the threshold value (£6.28 per test) MT would be dominated and
not a rational option. When the unit cost of pre-clinical testing is
only slightly lower then the threshold value MT is subject to
extended dominance [36]. This implies that even though it is less
costly than CSFS a more cost effective strategy could be obtained
by applying a mix of ST and CSFS which would have a lower cost
than MT for the same, average, number of infected sales averted.
This would require a mixed strategy to be technically and
politically feasible and makes the strong assumption that the
effectiveness scale of sales avoided maps linearly to societal
benefits. In other words, in terms of scrapie control and
eradication strategies, a mixed strategy using a combination of
genotype and pre-clinical testing in flocks may under certain test
unit cost conditions be the most cost-effective option. Although
this may be the most efficient allocation of resources, it may not be
the most fair or equitable [36]. Different sheep would be exposed
to different control measures (and outcomes) within flocks. We
have not considered sensitivity to the costs of genotyping as these
are well-established. However, should these change, for example if
a reduction in the volume of samples genotyped results in a
concomitant increase of per unit cost, this would potentially
increase the relative attractiveness of a pre-clinical test based
strategy. Similarly, we base this analysis on an average
compensation rate for sheep – some pedigree breeding stock are
dramatically more valuable than others, and therefore may result
in very different decision points if flocks are considered on an
individual basis. As losses to a farmer may exceed compensation
costs in the case of higher valued pedigree animals, the relative
costs of different strategies may vary. Specifically, the cost of the
CSFS strategy is highly influenced by compensation costs and may
make MT a potentially desirable strategy in high value flocks.
If there are no budget constraints, the CSFS would achieve the
most effective outcome and would be the most fairly and ethically
applied strategy across the population [36].
Of course the sensitivity of the pre-clinical test and the
proportion of susceptible genotypes in the flock will alter the
cost-effectiveness of different control strategies in high risk flocks.
As test sensitivity and proportions of susceptible genotypes
decreases, the cost of MT has to be increasingly cheaper than
current estimates for it to be considered a reasonable alternative to
the CSFS. If the prevalence of susceptible genotypes decreases
over time [37] as has also been seen in the Netherlands [38], in
high risk flocks the difference in the effectiveness between the ST
and MT/CSFS will become smaller. Under those conditions, the
ST could become an alternative to the MT or CSFS in the future.
At present, this analysis was simplified to only consider the costs
and effects of each strategy in the present (i.e. no discounting of
Figure 3. Cost-effectiveness plane examining all flocks (high, moderate and low risk) assuming 40% proportion of susceptible
genotypes under three preclinical test unit costs (0.5, 1.0 and 1.5 times threshold price that would give CSFS and MT equal costs).
The MT strategy is clearly dominated in the 1.5 times scenario having a much greater cost than CSFS with lower effectiveness (fewer infected sheep
sales avoided.).
doi:10.1371/journal.pone.0032884.g003
Economic Evaluation of Scrapie Testing Strategies
PLoS ONE | www.plosone.org 7 March 2012 | Volume 7 | Issue 3 | e32884

costs and effects was applied). A deterministic sensitivity analysis
was used to examine the impact of test sensitivity and proportion
of susceptible genotypes in the flock.
We have compared the efficacy of each strategy by comparing
the reduction in the number of infected sheep sold. However, we
recognize that there is potentially more than one measure of effect
for each strategy for each scenario. For example, from a
government/industry perspective, the risk of disease spread and
subsequent trade restrictions and animal welfare may be the most
important effects to measure the efficacy and ultimately effective-
ness of a successful control strategy. Alternatively, a farmer may
consider loss of performance traits, the effect of inbreeding, loss of
genetic diversity, susceptibility to other diseases and trade
restrictions more important. Equally, a societal perspective may
also take into account consumer interest in disease-free meat.
Measuring multiple effects that may occur during a scrapie
epidemic (such as trade and food security disruption and animal
welfare) and measured in the context of economic analyses may be
best represented by a cost-benefit analysis [2] and this will be
considered in future studies.
In this study, we have shown that extending a previous
epidemiological analysis [16] to consider economics presents
additional options that may have a considerable benefit to animal
health and welfare. Such combined epidemiological economic
analyses are in their infancy when considering infectious disease
dynamics, but are often important when considering policy advice
that must consider the complexities of nonlinear infectious disease
dynamics.
Author Contributions
Conceived and designed the experiments: LB IH RK. Performed the
experiments: LB IH. Analyzed the data: LB IH. Wrote the paper: LB IH
NH FH HF RK.
References
1. James A (2005) The state of veterinary epidemiology and economics. Prev Vet
Med 67: 91–99.
2. Drummond MF, Sculpher MJ, Torrance GW, O’Brien BJ, Stoddart GL (2005)
Methods for the economic evaluation of health care programmes. Third edition.
Oxford: Oxford University Press.
3. Bates TW, Carpenter TE, Thurmond MC (2003) Benefit-cost analysis of
vaccination and preemptive slaughter as a means of eradicating foot-and-mouth
disease Am J Vet Res 64(7): 805–812. doi:10.2460/ajvr.2003.64.805.
4. Berentsen PBM, Dijkhuisen AA, Oskam AJ (1992) A dynamic model for cost-
benefit analyses of foot-and-mouth disease control strategies Prev Vet Med 12:
227–243.
5. Disney WT, Green JW, Forsythe KW, Wiemers JF, Weber S (2001) Benefit-cost
analysis of animal identification for disease prevention and control OIE Revue
Scientifique et Technique 20: 385–405.
6. Miller GY, Tsai J-S, Forster DL (1996) Benefit-cost analysis of the national
pseudorabies virus eradication program JAVMA 208(2): 208–213.
7. Brisson M, Edmunds W (2003) Varicella vaccination in England and Wales: a
cost-utility analysis. Arch Dis Child 88: 862–869.
8. Caughey B, Chesebro B (1997) Prion protein and the transmissible spongiform
encephalopathies. Trends Cell Biol 7: 56–62.
9. Foster JD, Hope J, Fraser H (1993) Transmission of Bovine Spongiform
Encephalopathy to sheep and goats. Vet Rec 14: 339–341.
10. Bruce ME, Will RG, Ironside JW, McConnell I, Drummond D, et al. (1997)
Transmissions to mice indicate that ‘new variant’ CJD is caused by the BSE
agent. Nature 389: 498–501.
11. Hill AF, Desbruslais M, Joiner S, Sidle KCL, Gowland I, Collinge J, et al.
(1997a) The same prion strain causes vCJD and BSE. Nature 389: 448–450.
12. Hill AF, Zeidler M, Ironside J, Collinge J (1997b) Diagnosis of a new variant
Creutzfeld-Jakob disease by tonsil biopsy. The Lancet 349: 99–100.
13. Defra (2007) NSP update. Technical Report 9 (Spring/Summer). Defra 2007.
14. Kao RR, Gravenor MB, McLean AR (2001) Modelling the national scrapie
eradication programme in the UK. Mathematical Biosci 174: 61–76.
15. Baylis M, Houston F, Kao RR, McLean A, Hunter N, et al. (2002) BSE- a wolf
in sheep’s clothing? Trends Microbiol 12: 563–570.
16. Boden LA, Houston F, Fryer H, Kao RR (2010) Use of a Pre-clinical Test in the
Control of Classical Scrapie. J Gen Virol 91: 2642–2650. doi: 10.1099/
vir.0.022566-0.
17. EC2001/999 Available: ,http://eurex.europa.eu/Notice.do?val=259678:cs&lang
en&list=259678:cs,427264:cs,&pos=1&page=1&nbl=2&pgs=10&hwords=..
18. EC, 2003a EC, Commission decision of 13 February 2003 laying down the
minimum requirements for the establishment of breeding programmes for
resistance to transmissible spongiform encephalopathies in sheep (2003/100/
EC), Off J Eur Union L41: 41–45.
19. EC, 2003b EC, Commission Regulation (EC) No 1915/2003, Off J Eur Union
L283: 29–33.
20. Gonza´ lez L, Dalgleish MP, Martin S, Dexter G, Steele P, et al. (2008) Diagnosis
of preclinical scrapie in live sheep by the immunohistochemical examination of
rectal biopsies. Vet Rec 162: 397–403.
21. Stringer SM, Hunter N, Woolhouse MEJ (1998) A mathematical model of the
dynamics of scrapie in a sheep flock. Math Biosci 153: 79–98.
22. Hagenaars TJ, Donnelly CA, Ferguson NM, Anderson RM (2000) The
transmission dynamics of the aetiological agent of scrapie in a sheep flock. Math
Biosc 168: 117–135.
23. Fryer HR, Baylis M, Sivam K, McLean A (2007) Quantifying the risk from
ovine BSE and the impact of control strategies. Proc R Soc 274: 1497–1503.
24. Trusco tt JE, Ferguson NM (2009) Control of scrapie in the UK sheep
population. Epidemiol Infect 137: 775–86. doi: 10.1017/50950268808001064.
25. Gubbins S, Webb CR (2005) Simulation of the options for a national control
programme to eradicate scrapie from Great Britain. Prev Vet Med 69: 175–87.
26. Gubbins S, Webb CR (2005) Simulation of the options for a national control
programme to eradicate scrapie from Great Britain. Prev Vet Med 69: 175–187.
27. Durand B, Martinez M-J, Calavas D, Ducrot C (2009) Comparisons of strategies
for substantiating freedom from scrapie in a sheep flock. BMC Vet Res 5: 16. doi
10.1186/1746-6148-5-16.
28. Economic analysis of scrapie. Available: http://www.apd.reading.ac.uk/AgE-
con/livestockdisease/sheep/scrapie1.htm. Accessed 2011.
29. Gravenor MB, Cox DR, Hoinville LJ, Hoek A, Mclean AR (2001) The flock-to-
flock force of infection for scrapie in Britain. Proc Biol Sci 268(1467): 587–592.
30. McIntyre KM, Gubbins S, Goldmann W, Hunter N, Baylis M (2008)
Epidemiological Characteristics of Classical Scrapie Outbreaks in 30 Sheep
Flocks in the United Kingdom. PLoS ONE 3(12): e3994. doi:10.1371/
journal.pone.0003994.
31. Sivam SK, Baylis M, Gravenor MB, Gubbins S (2006) Descriptive analysis of an
anonymous postal survey of the occurrence of scrapie in Great Britain in 2002.
Vet Rec 158: 501–506.
32. Fediaevsky A, Calavas D, Gasqui P, Moazami-Goudarzi K, Laurent P, et al.
(2010) Quantitative estimation of genetic risk for atypical scrapie in French
sheep and potential consequences of the current breeding programme for
resistance to scrapie on the risk of atypical scrapie. Genetics Selection 42: 14.
doi: 10.1186/1297-9686-42-14.
33. Green D, del Ri Vilas VJ, Birch CPD, Johnson J, Kiss IZ, et al. (2007)
Demographic risk factors for classical and atypical scrapie in Great Britain J Gen
Virol 88: 12 3486–3492. doi: 10.1099/vir.0.83225-0.
34. Kao RR, Green DM, Johnson J, Kiss IZ (2007) Disease Dynamics over very
different timescales: FMD and scrapie on the Network of Livestock Movements
in the UK. J Roy Soc Interface 4: 907–916. doi:10.1098/rsif.2007.1129.
35. Fediaevsky A, Maurella C, No¨remark M, Ingravalle F, Thorgeirsdottir S, et al.
(2010) The prevalence of atypical scrapie in sheep from positive flocks is not
higher than in the general sheep population in 11 European countries BMC Vet
Res 6: 9. doi:10.1186/1746-6148-6-9.
36. Cantor SB (1994) Cost-Effectiveness Analysis, Extended Dominance, and Ethics:
A Quantitative Assessment Med Decis Making 14: 259. doi: 10.1177/
0272989X9401400308.
37. Arnold M, Meek C, Webb CR, Hoinville LJ (2002) Assessing the efficacy of a
ram-genotyping programme to reduce susceptibility to scrapie in Great Britain.
Prev Vet Med 56: 227–249.
38. Hagenaars TJ, Melchior MB, Bossers A, Davidse A, Engel B, et al. (2010)
Scrapie prevalence in sheep of susceptible genotype is declining in a population
subject to breeding for resistance. BMC Veterinary Research 6: 25. Available:
http://www.biomedcentral.com/1746-6148/6/25. Accessed 2011.
Economic Evaluation of Scrapie Testing Strategies
PLoS ONE | www.plosone.org 8 March 2012 | Volume 7 | Issue 3 | e32884