Example title

1
BAYESIAN HIERARCHICAL MODELS FOR THE EVALUATION
AND DESIGN OF SURVEILLANCE SYSTEMS AN
APPLICATION TO SCRAPIE SURVEILLANCE IN GB
Alberto Vidal Diez1, Mark Arnold1, Victor del Río
Vilas1,2 1 Centre of Epidemiology and Risk
Analysis,Veterinary Laboratories Agency, UK 2
Department for Enviroment, Food and Rural Affairs
2
WHAT IS SCRAPIE?

• Scrapie is a fatal, infectious neurological
  disease that occurs in sheep and goats. It is a
  transmissible spongiform encephalopathy (TSE). It
  presents a long incubation period. The disease
  occurs mostly in 2 to 5 years-old sheep.
• Funding for scrapie surveillance and eradication
  (millions of US Dollars)

Total 1301.380.000

• Main sources of active surveillance in GB
• -Fallen sheep over 18 months of age (FS)
• -Healthy slaughtered sheep over 18 months of age
  (AS)
• Compulsory Scrapie Flock Scheme (CSFS) is the
  programme to monitor the infected holdings.
• In 2002 it detected an atypical form of scrapie.

3
Objectives

• Estimate the sensitivity of the current
  surveillance system with respect to simulated
  true holding prevalence and study main factors
  which reduce Se.
• Investigate the possibility of stratifying the
  sampling by county or by holding size.
• Determine an optimal sample size at holding and
  at county level to improve the sensitivity of the
  surveillance.

4
Methodology
(Branscum et al,2006)
are the sensitivity and specificity of the
diagnostic test for the abattoir survey,
respectively.
and
5
Methodology
6
Assumptions

• We assumed 100 sensitivity and specificity for
  the screening test in both surveillance sources.
• We used only holdings with more than 20 adult
  sheep
• We assumed homogeneity of the prevalence across
  the country in the scenarios but the programme
  was implemented to model heterogeneity
• We used the binomial distribution
• Most of the scenarios assumed that the
  probability that an infected sheep is going to
  die on the farm is about 91
• We assumed that 5 sheep died on farm and 20
  sheep are sent to the abattoir each year.

7
(No Transcript)
8
First stage
9
(No Transcript)
10
Scenarios (Fallen Stock)
Differences between stratifications 1 TP
holding prevalence 0.45 AP holding prevalence
(Del Rio Vilas et al, 2008) 0.65 TP and AP
within-holding prevalence (Ortiz-Pelaez and Del
Rio Vilas, in press). 30 TP and AP
within-country prevalence. 91 of the infected
sheep die on the farm
Compare samplings which test small number of
sheep in each holding but testing many holdings
vs. more sheep within holding from a reduced
number of holdings
11
Scenarios (Fallen Stock)
Impact of assuming the within-holding prevalence
suggested by the European Union (2) 1 TP
holding prevalence 0.45 AP holding
prevalence 2 TP and AP within-holding
prevalence 91 of the infected sheep die on the
farm
Impact of the assumption that 91 of sheep die on
farm. 40 of the infected sheep die on farm.
12
Conclusions Further research

• These Bayesian Hierarchical Mixtures models are
  very flexible, useful and powerful to evaluate
  surveillance strategies. They can combine
  multiple sources of surveillance and complex
  sampling designs can be investigated. They can be
  used to review surveillance strategies each year
  to adapt the system to the evolution of the
  disease.
• Generate a model across time to design
  surveillance strategies for longer periods
• In the case of scrapie, we could model the
  effect of monitoring the detected holdings each
  year.
• We could investigate the effect of vaccination
  programmes.
• We could use network models with transmission
  rates to add the spreading of the disease every
  year .
• Add a function of costs
• Investigate the use of a backward method to
  estimate the
• true prevalence

13
Acknowledgements
Angel Ortiz-Pelaez, Robin Sayers, Colin Birch,
and Zoe Chapman
14
References

• - Branscum A.J., Johnson W.O, Gardner,I.A.(2006)
  Sample size calculations for disease freedom and
  prevalence estimation surveys. Statistics in
  Medicine 252678-2674
• Del Rio Vilas, V.J., Bohning, D. (2008).
  Application of one-list capture-recapture models
  to scrapie surveillance data in Great Britain.
  Prev Vet Med 85 253-266.
• - Ortiz Pelaez, A., Del Rio Vilas, V.J. (in
  press). Within-holding prevalence of sheep
  classical scrapie in Great Britain. BMC Vet Res