Title: Enhanced Surveillance for Avian Influenza and Other Disease Risks
1Enhanced Surveillance for Avian Influenza and
Other Disease Risks
- Professor Roger Morris and colleagues
- Massey University EpiCentre
- Palmerston North New Zealand
2Why do surveillance?
- To provide evidence that a disease is absent
- To reduce the risk of taking the wrong disease
control action - And the risk of failing to act when necessary to
do so - We are developing economic methods to decide how
to invest each dollar of surveillance money most
wisely
3Adapting surveillance to new needs
- Surveillance systems have not adapted fast enough
to changing requirements - Need to make surveillance more risk-based
- Diseases form a limited number of epitypes
require similar surveillance strategies - Scanning and targeted surveillance
- Design better systems to collect and analyse
surveillance data so that it is used effectively
for making wise decisions - Make outputs more current and targeted to the
various users of the data
4Risk-based surveillance
- Develop surveillance portfolio
- Balanced combination of techniques
- Scanning surveillance broad assessment
- Targeted surveillance answer specific questions
about a disease - Use information on risk factors for a disease to
help allocate effort - Develop risk landscape to guide decisions on
surveillance strategy surveillance data lowers
the risk mountains to a flat plain
5(No Transcript)
6The epidemiological toolbox
- Data gathering which can yield a valid and
insightful assessment of field disease situation - Analytical tools which can achieve this
- Modelling tools to evaluate and predict
- Surveillance portfolio to guide decisions
- Integrating information system to make it happen
7Disease mapping for surveillance
- GIS software now very important to interpret
patterns of disease - Especially if used jointly with other
epidemiological methods - Allows more cost-effective selection of samples
to include in surveillance system - Now have powerful analytical techniques to
interpret spatial patterns and relate to risk
factors
8(No Transcript)
9Prevalence of brucellosis in Armenia
10Data gathering methods
- Expanding range of methods used
- Participatory methods
- Syndromic surveillance
- Field investigation strategies
- Molecular epidemiology linked with other tools
- Laboratory data should be tightly targeted
- Integrate laboratory and field methods
- Multiple imperfect methods better than single
perfect source
11Participatory Disease Surveillance
12HPAI detections by PDS
13Thailand
14Farmer surveys simple but effective
300 duck farmers in southern Vietnam interviewed
in two weeks Are ducks really the problem here?
98 have more than 80 of ducks vaccinated But
60 of these people have less than 80 of
chickens vaccinated Ducks are not moved nearly as
far as senior vets thought
15AI investigation in villages and markets
- Investigating low path viruses as well as H5N1
can help give a better understanding of methods
of spread of AI viruses, and how to control them - In Hong Kong markets, spread involved drinking
water, minor poultry species (quail), pet rooster - Spread of H5N1 to local farms was from markets,
not the other way - Molecular epidemiology showed that there were
four incursions not one, with different spread
16Studies in Vietnam
- Vietnam has had a lot of outbreaks, has made good
progress in reducing the number per year - Analysed spatial and temporal pattern of poultry
outbreaks 2004-7, especially in two deltas - Association between human and poultry cases
- Role of ducks and other birds in AI spread,
factors influencing disease occurrence - Will use this information to develop a targeted
surveillance system, with computer tools to
support it
17Poisson regression of risk areas
- Examined pattern of outbreaks 2003-4 using a grid
overlaid on country - Areas with over 66 of land irrigated at much
higher risk of outbreaks - Areas below 250m altitude at increased risk
- Other factors examined were not influential
- Potential use of satellite imaging to guide
surveillance investment
18Red River delta
19Mekong River delta
20Spatial SatScan of clusters in RR delta
21Spatial autocorrelation in RR delta
22Spatial pattern repeats over years
23Human and poultry cases in Vietnam
a)
b)
24Temporal pattern of human and poultry cases
25How should we investigate AI?
- Sero-monitoring has only limited value in a
vaccinated population - Is NOT a disease surveillance method in this
situation, only checks vaccine coverage - Want 80 birds protected for AI
- To test whether protection level adequate, test
250 animals for any size population - So design sampling plan to test that vaccination
program is working in different sectors (3, 4),
production systems and regions of the country
26Disease surveillance
- Want to determine distribution of infection and
transmission patterns - PCR is a valuable technique, but only useful if
use virus isolation as well, to confirm that
virus is circulating and check PCR accuracy - PCR should always give some positives, even if no
infection present! - Virus isolation and molecular strain
investigation most useful investigation method to
understand virus transmission
27Investigating AI
- Focus should be on understanding how infection is
transmitted and maintained, then monitoring it - Main emphasis should be on investigating sector 3
and sector 4 flocks, live bird markets - Need epidemiological investigation strategy to
answer key questions through good design - If you ask the right questions, spend far less on
testing, but get more useful results - Use modern analytical methods
28Data Analysis
- Many powerful analytical methods now available
- Do not need perfect data provided that study
design is sound - Carefully designed intensive epidemiological
investigations to examine risk factors are far
more useful than large-scale sampling done
without collecting risk factor data - Then can use modelling and other techniques to
test various explanations of the disease pattern
29Number of cattle moved OUT of departments for
finishing Jan - Mar
Social network analysis can be used to
investigate disease spread pathways
Grey lines indicate movement of gt 1000 head for
3-month period. Symbols proportional to number of
cattle moved.
30Spatial modelling of disease
- Valuable method of assessing possible causes
- Valuable for evaluating control options
- We use generic spatial models which can be given
the profile of any known (or new) disease - Modelled foot and mouth disease epidemic in 2001
for Britain, modelling avian influenza with same
model
31(No Transcript)
32(No Transcript)
33Monitoring control program results
- Need baseline data on the situation before
control activities are started - Need to use a monitoring system which accurately
measures change in prevalence of disease and/or
infection - Need to gather information which will allow you
to detect and respond to weaknesses in the program
34Brucellosis control in Tajikstan
35Animal health information system
- Use an approach which suits the needs of the
country, start simple but with development plan - It must allow data analysis from the start, and
be able to advance as the countrys needs change - Should include both disease control and
surveillance - Ability to map disease data should be seen as an
important feature which must be available, though
it may not be used at first
36Developing a surveillance portfolio
- Retirement planning
- Invest in multiple assets to spread your risk
- Evaluate performance by expected return
- Consider variation around return
- Surveillance assets
- Invest in multiple surveillance techniques to
give balanced picture - Can use points per sample or investigation,
instead of - Each component of surveillance program has a
different cost and different number of points
achieved by particular test or investigation - Applied to BSE and Trichinella, adapting to other
diseases - Risk-based sampling can give better assurance of
disease state
37Risk-based sampling for exotic disease
Comparison of the number of tests allocated per
SA and the maximum possible disease prevalence in
ewe flocks (a 0.05) when distributing samples
using portfolio theory (PT) and proportional
allocation (PA)
38Risk assessment
Activate
P1
P2
P3
39Feedback to participants
- A major issue in achieving effective surveillance
is getting cooperation from data providers - Essential for completeness, and to minimise bias
- Feedback is most important stimulus to continuing
cooperation
40Conclusion
- We need to be ready to face and solve new disease
challenges - Effective surveillance is the key to quick
detection and effective control - Surveillance needs multiple sources of
information, and smart tools for interpretation - Need integrated surveillance and response
strategy if we are to control future diseases
successfully, and a toolbox of techniques to
quickly determine what is going on