OS16 - Open Session 2016 Cascais, Portugal 26 - 28 /10/2016 EuFMD Sessions\Open Session\Archive-2018\Open 2016 Cascais- Portugal\PPT presentations\

1. Open Session of the EuFMD - Cascais –Portugal 26-28 October 2016
PRIORITISATION OF RESOURCES
FOR EARLY DETECTION OF
DISEASE INCURSIONS
A Cameron, AM Bouhbal, I Ahamjik,
H Haj Ammar, K Ouali, M Khayli, F Rosso

2. Background
• FMD outbreak in North Africa
– 2014 and 2015
– CaHle and small ruminants

3. Overview
• FMD Surveillance
• Quality of early detecNon surveillance
• Early detecNon – risk components
• Economic model – theoreNcal
• PrioriNsaNon model – pracNcal

4. CLASSIFICATION OF SURVEILLANCE

5. ClassificaNon of purposes of surveillance
for FMD
• FMD present
– Measure the disease
– Case finding
• FMD absent
– Demonstrate freedom
– Early detecNon

6. Early detecNon surveillance
• How much should we do?
• How much should we spend?
• InternaNonal guidelines

7. Guidelines
• OIE
– Establishes requirement for countries to have an
early detecNon system
– Notes that disease reporNng and non-random
surveillance can be used for early detecNon
OIE Terrestrial Code, Chapter 1.4, Animal Health Surveillance

8. Guidelines
• FAO
– “Surveillance has as its main purpose, early
detecNon of disease.”
– MenNon that passive farmer reporNng can help
FAO Manual on Livestock Disease Surveillance and InformaNon Systems

9. Early detecNon surveillance
• We all agree that it is important and we must
do it
• Very liHle guidance available on how to do it
• Very liHle guidance on how much to invest

10. QUALITY STANDARDS

11. Quality standards for surveillance
• QuesNons people ask epidemiologists
– What sample size should I use?
– How much will it cost?
• QuesNons epidemiologists ask people
– How good do you want the result of your surveillance
to be?
– How much money have you got?

12. Quality standards…
• Measuring the level of disease
– Examples
• Seroprevalence
• Incidence
– Quality standards
• Precision
– Width of the 95% confidence interval
• Bias

13. Quality standards…
• Case finding
– Example
• Disease eradicaNon
– Quality standard
• What proporNon of cases do you want to find?
• Surveillance sensiNvity
• Can you eradicate disease if your sensiNvity is less than 100%?

14. Quality standards…
• DemonstraNon of freedom
– Example
• MeeNng standards for internaNonal organisaNons or trading partners
• Confirming that an eradicaNon program has been successful
– Quality standards
• Design prevalence
• Surveillance sensiNvity
• Confidence in freedom

15. Quality standards…
• Early detecNon
– DetecNon sensiNvity
• What proporNon of disease incursions do we want to
detect?
– DetecNon delay
• How quickly do we want to detect the disease?

16. From standards to pracNce
• Target quality standards for early detecNon
– Detect 100% of incursions
– Before they spread from the iniNal epidemiological unit
• ImplicaNons
– What proporNon of the populaNon needs to be under
surveillance?
• 100%
– How oben do they need to be observed/tested?
• All the Nme

17. Available approaches
• What surveillance systems do we have
that can achieve this?
– Farmer reporNng
• How effecNve if farmer reporNng?
– Variable…

18. Conclusions
• The only pracNcal tool we have for early
detecNon is imperfect
• What do we do to improve it?
– Invest
– How much and where?

19. Investment in surveillance
• Improved farmer reporNng
– Awareness, CommunicaNon, Training
• Supplement farmer reporNng with other surveillance
acNviNes
– AcNve clinical surveillance
– AbaHoir surveillance
– SenNnel surveillance
– Market or other aggregaNon points
– etc.

20. RISK-BASED SURVEILLANCE

21. Risk-based surveillance
• Risk-based surveillance
– Purpose
• BeHer surveillance at a lower costs
– Approach
• IdenNficaNon of sub-populaNons at higher risk
• Focusing surveillance resources on those higher risk
populaNons

22. ApplicaNons of risk-based surveillance
• DemonstraNon of freedom
– Yes, target high risk populaNons
• Case finding
– Yes
• Prevalence esNmaNon
– NO! Not representaNve à biased esNmates

23. QuesNon
• Can we use risk-based surveillance for
early detecNon?

24. The problem with early detecNon
• ProbabilisNc methods
– Based on repeated random events
with an underlying probability
– If it is repeated, we can predict the
future
• Disease incursions
– One-off events

25. General rule
• Risk-based surveillance that targets only the
high risk populaNon is not suitable for early
detecNon
• But, risk-based approaches can be used to
improve or supplement an exisNng farmer
reporNng system

26. QuesNon
• If we invest in early detecNon, what
should we be aiming to achieve?

27. Variability
• in sensiNvity of farmer reporNng system
– Invest in areas where it is weak
– Try to achieve uniformly high sensiNvity

28. Variability…
• in risk of introducNon
– Invest in areas of high risk

29. Variability…
• in consequences of an outbreak
– invest in areas of high consequence
• How do we prioriNse investment based
on the variability?

30. Risk components
• Probability of disease incursion
• Probability that a sub-populaNon will be
infected (relaNve to other sub-populaNons)
• Probability that the exisNng surveillance system
will fail to detect the incursion rapidly
• Consequences of failure to detect the incursion

31. Combining the factors
• 𝐸(𝑂𝐵)= 𝑃(𝑖𝑛𝑡𝑟𝑜)×(1− 𝑆𝑒)× 𝐶𝑜𝑛𝑠
Pr Pr $$
• Can this tell us how much to spend?
• Absolute risk, absolute cost

33. Problems
• TheoreNcal
– Hard to calculate or esNmate the true
consequences, or the true cost of prevenNon
– Budgets are oben not decided on the basis of need,
but availability

34. More pracNcal quesNon
• Given the available budget, where is it
best invested?
• ​ 𝐸( 𝑂𝐵)↓𝑖 =​ 𝑃(𝑖𝑛𝑡𝑟𝑜)↓𝑖 ×​(1− 𝑆𝑒↓𝑖 )×​
𝐶𝑜𝑛𝑠↓𝑖

35. Easy…

36. EsNmaNng parameters
• Absolute risk, absolute consequences
– Very difficult
• RelaNve risk, relaNve consequences
– Much easier

37. Example
• RR of introducNon in high movement area: 3
– Compared to low movement
• RR of detecNon failure in remote area: 4
– Compared to area with good vet services
• RelaNve consequences for breeder farm: 5
– Compared to farm that sends only to slaughter

38. Risk esNmates
• High movement, remote, breeder:
– 3 x 4 x 5 = 60
• High movement, local, slaughter
– 3 x 1 x 1 = 3
• Low movement, remote, breeder
– 1 x 4 x 5 = 20

39. OS16
IntroducNon
High Risk
Low Risk
Low SensiNvity
High SensiNvity
Low SensiNvity
High SensiNvity
Surveillance Consequences
High
Low
High
Low
High
Low
High
Low
50%
10%
13%
3%
17%
3%
4%
1%

40. Possible intervenNons
• Decrease risk of introducNon
• Increase sensiNvity of detecNon
• Decrease consequences

41. Investment
• Investment jusNfied if
– It actually impacts on the risks

42. Conclusions
• Risk analysis
– Can be used to esNmate maximum jusNfiable investment
• Early detecNon surveillance
– Must be considered as a component of a broader disease prevenNon system
• Farmer reporNng system
– FoundaNon of effecNve early detecNon
– Many ways to improve
• Simple risk esNmates
– can inform prioriNsaNon of investment
• Risk can be addressed by
– PrevenNng entry
– Improving surveillance
– Decreasing consequences

43. Thank you