The High Plains Project The High Plains Project Michael Ward, Professor of Epidemiology, Univ of Sydney Bo Norby, Assistant Professor of Epidemiology Bruce McCarl, DIstinguished Professor, Agricultural Economics Levan Elbakidze, Asst. Prof, Idaho Texas A&M University 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 W eek
Transcript
Slide 1
The High Plains Project Michael Ward, Professor of
Epidemiology, Univ of Sydney Bo Norby, Assistant Professor of
Epidemiology Bruce McCarl, DIstinguished Professor, Agricultural
Economics Levan Elbakidze, Asst. Prof, Idaho Texas A&M
University
Slide 2
The High Plains Project Goal: Develop and apply a decision
support system that assists policy formulation to efficiently
detect and respond to incursions of FMD within the high plains of
Texas develop framework for integrated decision support systems
timeframe: January to September, 2006
Slide 3
Deliverables 1.Realistic regional FMD epidemic model in an
integrated system 2.Decision support system for FMD incursion
assessment 3.Evaluation of mitigation strategies
Slide 4
Slide 5
8 counties 7,942 miles 2 92 feedlots 2,231,300 cattle on feed
411,019 grazing cattle 17,471 land parcels
Slide 6
Specific Objectives modify epidemic model structure to
represent livestock industry in the high plains 3 months integrate
epidemic model with economic models 6 months investigate the
consequences of FMD incursions 1 month determine the impact of a
range of mitigation strategies 2 months
Slide 7
Identify Key Industry Components Form spatial, herd type
representation of High Plains animals Estimate herd to Herd Animal
and Other Contact Rates Setup linked regional Epidemiologic/
Economic Model Study consequences of alternative Disease
introduction scenarios Study consequences of ex ante and ex post
interventions Stake holder reporting Follow on Legend Done First
cut done, detailed underway Nearing completion Just beginning Study
Design
Slide 8
Animal and Contact Data: The High Plains Project Bo Norby,
Assistant Professor of Epidemiology National Center for Foreign and
Zoonotic Disease Defense Texas A&M University
Slide 9
High Plains Study Area Interviews as per August 28, 2006
Operation typesDoneScheduled Cattle on feed30 Dairies121
Swine27
Slide 10
30 Feedlots Interviewed as of August 28, 2006 High Plains Study
Area August 3, 2006
Slide 11
FeedlotsAverageMedianMinimumMaximum
Capacities2529,84827,1002,40075,000 Cattle received Cattle received
per month 254,0934,3003225,500 Locations received from per day
252.72.41.07.5 Days per month receiving cattle 25119322 Average
direct contacts per day 0.99 Cattle shipped Cattle shipped per
month 254,5564,00030010,000 Locations shipped to per day 252.4215
Days per month shipping cattle 251110225 Average direct contacts
per day 0.8 High Plains Study Area
Slide 12
Slide 13
Application of Epidemic Models: The High Plains Project Michael
Ward, Professor of Epidemiology National Center for Foreign and
Zoonotic Disease Defense Texas A&M University
Slide 14
can cause economically devastating outbreaks in countries free
of infection $US 1.6 billion, Taiwan 1997 (Yang et al., 1997) >8
billion, United Kingdom 2001 (Kao, 2003) estimated costs of
incursions $AUD 213 billion, Australia (Anon., 2002) $US 6 billion,
California (Ekboir, 1999) $US 14 billion, United States (Paarlberg
et al., 2002) Foot-and-Mouth Disease
Slide 15
response to incursions: rapid application of appropriate
control measures decisions with little current or empirical data
political, economic and other issues disease spread models:
guidance on probable extent and time span of an outbreak design of
optimal control strategies policy formulation gap analysis:
research agenda links between disease modelers, stakeholders,
decision-makers FMD Models
Slide 16
state transition, stochastic model spatial distribution of
livestock industry contact structures default herd types FMD virus
transmission options sale barns, auctions, order buyers wind-borne
spread spread rate parameter AusSpread Model
Slide 17
AusSpread(Garner and Beckett, 2005) SLIR: Susceptible - Latent
- Infectious Recovered Stochastic state transition model which
operates within a geographic information system (GIS) framework
Probabilities of transition functions of direct and indirect
contacts sale barns, order buyers and windborne spread Susceptible
LatentSymptomatic Recovered/ Removed / dead Infection rate - Rate
at which latent become infectious Rate of recovery
High Plains Study: feedlots specific location of some feedlot
types unknown feedlot locations with capacity estimates known
allocated by size: 1.>50,000 feedlot 1, company-owned 2.20,000
50,000 feedlot 2, stockholder 3.5,000 20,000 feedlot 3, custom
4.feedlot 4, backgrounder: known locations 5.
High Plains Study: slaughter slaughter begins 1 day post
detection (day 15); saleyards shut day 15: only 1 herd slaughtered
(index case) days 16-24: 5 herds/day slaughtered days 25-30: 7
herds/day slaughtered days >30 10 herds/day slaughtered
probability of each targeted herd being slaughtered per day:
(capacity for the day) (total number of herds to be
slaughtered)
Slide 26
High Plains Study: vaccination vaccine unavailable until day 21
days 21-28:12 herds/day vaccinated days 29-34:25 herds/day
vaccinated days >3450 herds/day vaccinated probability of each
herd being vaccinated per day: (capacity for the day) (number of
herds awaiting vaccination) vaccination reduces the resources
available for slaughter by 25% dangerous contact (DC) slaughter
stops once vaccination begins (backlogged DC herds placed under
surveillance) suspect herds, DC and dead herds not vaccinated
Slide 27
High Plains Study: current status 100 model runs per
scenario/mitigation combination run time: ~12 hours ~12 days
current status: ScenarioSlaughterVaccination backyard saleyard
small feedyard large feedyard = currently running
Slide 28
Slide 29
uncontrollable outbreaks (>10% of herds infected by day 60
120): ScenarioSlaughterVaccination backyard 00 saleyard 100 small
feedyard 410 large feedyard 2951 High Plains: preliminary
results
Slide 30
FMD incursions: slaughter only mitigation strategy Herds
infected Scenario Av. herds infectedMinimumMaximum Backyard6135
Small feedyard642459 Large feedyard18361561
Slide 31
FMD incursions: slaughter vs. slaughter and vaccination Herds
infected
Slide 32
FMD incursions: slaughter only mitigation strategy Epidemic
length Scenario Av. epidemic lengthMinimumMaximum Backyard282145
Small feedyard534118 Large feedyard754788
Slide 33
FMD incursions: slaughter vs. slaughter and vaccination
Epidemic length
Slide 34
Number of herds infected, small feedlot introduction Herds
infected
Slide 35
Number of herds infected, small feedlot introduction Cattle
slaughtered
Slide 36
FMD incursions: small feedlot scenario Total cattle slaughtered
Slaughter only Slaughter and vaccination Ratio
Average18,03015,8211.14 Minimum8498251.03 Maximum119,72568,3421.75
IQR 2,803 26,2992,416 68,3421.16 2.60 interquartile range (IQR):
50% of simulation results
Slide 37
FMD incursions: small feedlot scenario Daily average number of
cattle slaughtered Slaughter only Slaughter and vaccination Ratio
Average3603001.20 Maximum2,0362,0041.02 IQR52 55022 4452.36
1.24
Slide 38
The next steps technical issues data needs and expert opinion
geographic distribution of feedlot types update other herd types
contact rates: are these realistic? total contacts versus adequate
contacts: have we seriously overestimated spread from
feedlots?
Slide 39
Contacts=total number of contacts per day; DContact=total
number of direct contacts per day; IContact=total number of
indirect contacts per day; PIDC=probability of infection from
direct contact; PIIC=probability of infection from indirect contact
Direct and Indirect Contacts
Slide 40
The next steps technical issues data needs and expert opinion
number of buyers at sales, destination of purchased cattle
probability of buying latently infected animals at a sale 20%
likely to be too high probability of selling cattle from an
infected herd via a sale 20% likely too high epidemic-economic
link: automated carcass disposal and capacity link transportation
model for long-distance spread faster simulation speed: re-coding,
grid computing
Slide 41
The next steps Vulnerability assessment critical components of
the system E.g. Feed availablity the role of sales, order buyers
feedlot vulnerabilities seasonality multiple FMD incursions
wildlife reservoirs?
Slide 42
The next steps Policy issues increase range, diversity of
scenarios simulated what are the important questions? increase
types of mitigation strategies simulated what are the options?
policy and response implications how do you use insights gained to
assist policy formulation and decision-making?
Slide 43
The next steps Key decision issues time-to-detection
vaccination strategies limited labor availability welfare slaughter
and feed availability other vaccine strategies e.g.blanket (area)
vaccination vaccination to live vs. kill targeted vaccination time
to availability of vaccination, decision to employ use of
anti-virals timing of control strategies
Slide 44
High Plains Study design Identify Key Industry Components Form
spatial, herd type Representation of High Plains animals Estimate
herd to Herd Animal and Other Contact Rates Setup linked regional
Epidemiologic/ Economic Model Study consequences of alternative
Disease introduction scenarios Study consequences of ex ante and ex
post interventions Stake holder reporting Follow on Legend Done
First cut done, detailed underway Nearing completion Just
beginning
Slide 45
Economic Effects: The High Plains Project Bruce McCarl, Regents
Professor, Ag. Economics National Center for Foreign and Zoonotic
Disease Defense Texas A&M University
Slide 46
46 Economic Valuation Module Estimate losses to local livestock
industry Costing Module Epidemiologic output Valuation Assumptions
Results Data
Slide 47
47 Economic Module Economic costing of an event Livestock cost
components Value of lost animals Lost gross income (60 days) Value
of vaccinated animals: slaughtered or salvaged (50% value loss)
Carcass disposal Quarantined animals Welfare cost/slaughter Cost of
strategy implementation Value of alternative disease management
policies Quarantines Vaccination Ring and Target Culling/euthanasia
Animal Traceability Surveillance Disinfection Disposal
Vulnerability Assessment Applying economics to the problem of
balancing prevention, preparedness and response under limited
budget Incentive policy design Market Effects Applied to a random
set of outbreak simulations
Slide 48
48 Economic Epidemiological link Linked econ model handles
Valuation of lost herds according to distribution by 21 animal
categories Gender Age Weight Type Gross income associated with lost
animals Gross income associated with periods of inactivity
Differential value of alternative herds by size and type Costing of
strategies by herd size Welfare slaughter Does this do the market
price and consumers surplus?
Slide 49
49 Data Spread of disease an effects of control options Epi
model ARS (USDA) Premise boundary data Texas Commission on
Environmental Quality data on CAFOs from face-to-face interviews
with producers (34 feedlot, 21 dairy, and 16 swine were
interviewed) - Management practices and direct and indirect contact
rates Costs of Control Options Schoenbaum and Disney 2003,
Carpenter and/or Bates Industry representatives Market values
Extension Specialists USDA publications, Texas Agricultural
Statistics Industry representatives Herd Compositions Interviews
Extension specialists Various operation costs and assumptions
Industry representatives
Slide 50
Data Spread of disease an effects of control options Epi model
Costs of Control Options Schoenbaum and Disney 2003, Carpenter
and/or Bates Market values Extension Specialists USDA publications,
Texas Agricultural Statistics Herd Compositions Surveys (Bo)
Extension specialists Various operation costs and assumptions
Industry representatives
Slide 51
Welfare slaughter Data from the 2001 UK FMD outbreak reveal
that 39% of animals slaughtered for welfare purposes because of
movement bans and limited feeding capacity. There is cost saving
potential if welfare slaughter can be reduced. High plains is at
risk 3-6 days feed supply Quarantine longer than that Mature/young
animals may need to be held plus in transit We examined assumption
that in a 3 KM ring there would be no movement and after 8 days in
that zone intensive operation cattle lost unless feed left in
Slide 52
Scenarios ScenarioIntroductionStrategy 1BackyardSurveillance,
Quarantine (3 kilometers), Slaughter of infected and dangerous
contacts 2 + target vaccination 3SaleyardSurveillance, Quarantine,
Slaughter of infected and dangerous contacts 4 + target vaccination
(still running) 5Small FeedlotSurveillance, Quarantine, Slaughter
of infected and dangerous contacts 6 + target vaccination 7Large
feedlotSurveillance, Quarantine, Slaughter of infected and
dangerous contacts 8 + target vaccination
Slide 53
Bottom Line Summary of Average Event Costs by Control Strategy
in Million $ Back yard Sale Yard Small Feed Lot Large Feed Lot
W/out Feed W/out vaccination 1101215172415 With vaccination
39314503 Value vaccination 71-142-88 With Feed W/out vaccination
7121295355 With vaccination 4314503 Value vaccination 3-219-148
Value Feed W/out vaccination 10337760 With vaccination 3500
Slide 54
Risk Results what do we see 100 trials each with prob 0.01 Max
amount of money lost with a probability level 10% of time lose
$210,000 or less 50% of time $85,650,000 or less 99% of time
$345,800,000 or less Prob 0.10.21 0.2515.66 0.585.65 0.75184.66
0.9206.87 0.95240.60 0.99345.81 Avg110.30
Slide 55
Risk Results Feed/No Vaccination Gain to feed with a
probability level 10% of time $0 or less 50% of time $83,300,000 or
less 99% of time $270,000,000 or less Moving line to upper left is
best NO Feed YES Feed 0.100.2 0.2515.70.6 0.5085.62.3 0.75184.73.0
0.90206.913.1 0.95240.643.6 0.99345.875.9 Avg110.339.0
Slide 56
Results on all strategies Vaccinate and feed the best Returns
to vaccinate much lower with feed Returns to feed lowered by
vaccinate Risk much lower with changed managment
Slide 57
Results on all strategies No Vaccinate and feed the best
Returns to vaccinate negative Returns to feed not very great Risk
great better management alternative?
Slide 58
Results on all strategies No Vaccinate and feed the best
Returns to vaccinate negative Returns to feed not very great Risk
great better management alternative?
Slide 59
Results on all strategies Feed slightly the best No vaccinate
results yet Returns to feed not very great Risk great cant hardly
manage at all? better alternatives?
Slide 60
Animal Disease Control Options and Objectives Objectives Early
detection vs. late detection Day 7 vs. day 14 (Ward et al. 2007 )
Early vs. late vaccine availability Day of detection vs. +7 Ring
(emergency, 5 km) and target (protective) vaccination Regular vs.
enhanced surveillance 3 herds on 1 st day and suspects twice/week
vs. 6 herds on 1 st day and suspects four/week Culling 1 day post
detection 1 to 10 on day 21
Slide 61
61 Mitigation: Large Fdlt. Backgrounder Fdlt Lrg. beefBackyard
Ring Depopulation Early detection1234 Late detection5678 Adequate
vaccine9101112 Inadequate vaccine13141516 Enhanced Surveillance
Early detection17181920 Late detection21222324 Adequate
vaccine25262728 Inadequate vaccine29303132 Ring Vaccination Early
detection33343536 Late detection37383940 Adequate vaccine41424344
Inadequate vaccine45464748 Targeted Vaccination Early
detection49505152 Late detection53545556 Adequate vaccine57585960
Inadequate vaccine61626364 Scenarios
Slide 62
Results Early detection 62 0 50 100 150 200 250 Large
FeedlotBackgrounder Feedlot Large BeefBackyard Type of Herds Median
Economic Costs Early Detection Late Detection
Slide 63
Results Early vs. late Vaccine availability 63 0 50 100 150 200
250 300 350 Large FeedlotBackgrounder Feedlot Large BeefBackyard
Type of Herds Adequate Vac. Inadequate Vac. Median Economic
Costs
Slide 64
Results Enhanced surveillance 64 0 20 40 60 80 100 120 140 160
180 200 Large FeedlotBackgrounder Feedlot Large BeefBackyard Type
of Herds Median Economic Costs Enhanced Surveillance Regular
Surveillance
Slide 65
Results Early detection reduced the median epidemic costs by
$150 million (68%)- Large Feedlot, $40 million (69%)- Backgrounder
Feedlot, $5 million (74%) - Large Grazing, $3 million (97%) -
Backyard introductions Vaccine availability and early application
Increased total costs dues to costs of vaccination and assumed 50%
loss in animal value Enhanced surveillance Increased by $53 million
(45%) -Large Feedlot. Decreased by $16 million (31%) - backgrounder
feedlots, Decreased $1 million (23%)- large grazing operations,
Decreased $1.6 million (77%) - backyard incursion scenarios,
Slide 66
Results: CDFs of losses in $ Millions 66
Slide 67
Results Break even RAC (McCarl 1990) Large feedlot RAC (0.01;
0.09) early detection enhanced surveillance (scenario 17) Otherwise
regular surveillance and early detection (scenario 1) Backgrounder
feedlot RAC -0.09 early detection and regular surveillance
(scenario 12) Large grazing RAC < 0.13 regular surveillance and
early detection (19) RAC > 0.13 enhanced surveillance and early
detection (3) Backyard For all RAC Enhanced surveillance and early
detection (20) 67
Slide 68
Results Welfare cost depends on introduction scenario High cost
in high density and multiple introduction Feed truck disinfection
leads to gains not always large Vaccination Effective under
backyard introduction Ineffective under feedlot introductions
Slide 69
Next Econ Activities in High Plains Study More on welfare
slaughter Formulation of strategies to examine More on vaccination
Preventative Feed Truck disinfection Sanitary practices Other?
Slide 70
Follow on Econ Activities in High Plains and more General FAZD
Follow on Balance Problem Examine risks that merit select
strategies Examine resiliency / robustness Examine compensation and
cooperation Add strategies Carcass disposal Disease mkt events
Slide 71
Scope of Strategy Manipulations Size of slaughter and
vaccination ring Speed of detection Diagnostics Animal inspection,
sensors? Contact rates Pre event Post event discovery time to fall
Resources for control Slaughter force Vaccination force Timing and
quantity of dose availability Feed and welfare slaughter
Alternative vaccination strategies Anti-virals Carcass disposal and
slow down strategies